SMPS opto feedback revolutionised

[cupoftea]

Hi,
Why is it that the attached opto feedback loop circuit is not more common?

The delivery of the opto transistor current into the emitter of a PNP is seen as a very high impedance, and as such, the effect of the "nuisance" optocoupler pole is mostly cancelled out......meaning you can compensate more like you would if there were no opto isolation.

Much higher feedback loop crossovers are easily possible, and so therefore, much improved transient performance is easily delivered.

Also, the NPN amplifies the signal current such that much lower opto current is necessary, leaving you less vulnerable to degradation of opto CTR with age and temperature.

LTspice and jpeg as attached

 

[mtwieg]

It's called a cascode.

 

[FvM]

Unfortunately not fast...

 

[crutschow]

The question is, are the added parts worth the supposed improvement in performance of the regulator?

 

[dick_freebird]

Offhand I don't see how speed is improved. Current is reduced
(alpha) and capacitance is added. Maybe some minor changes
to 5V power-up behavior.

Maybe you want to run side by side versions, with and without
the gingerbread, and see whether there's an actual speed
improvement (or any improvement in any aspect) and what
"fresh and interesting behaviors" it may add.

I would expect the PNP.C / NPN.B node to be real soggy in the
turnoff direction. Not at all fast, there, with no base discharge path.
Also don't understand the assertion about PNP.E being a high-Z
port, that is only true (-ish) because the NPN.B on the other side
is low (normal) current. But impedance is not increased by the
PNP, only passed through (divided by alpha).

 

[Easy peasy]

there are better extant ways to drive an opto .....