SMPS optocoupler feedback improved with common base BJT

[treez]

Hello,

Fig 13, page 9 of this app note shows an improvement on the "normal" optocoupler feedback connection for transformer isolated SMPS's.....

https://www.onsemi.com/pub_link/Collateral/AND8273-D.PDF


....On page 9 it says that the opto diode is driven by a current source.
Does this configuration, with the extra BJT, Q1, absoluteley need have to have a current source to drive the opto diode?

What would happen if i drove the opto diode with a typical TL431 circuit as in the following......

(please see schematic on bottom right of page 5 of the following)
https://www.onsemi.com/pub_link/Collateral/TND381-D.PDF

Also, since using the BJT, Q1, in fig 13 of the first app note gets rid of the opto-diode pole problem, why is this method not more frequently used?

 

[mtwieg]

The cascode and the current source drive both help, but they aren't both necessary. You can use the cascode without the current source drive, but you just won't get the benefit of the current source.

Also, since using the BJT, Q1, in fig 13 of the first app note gets rid of the opto-diode pole problem, why is this method not more frequently used?

Because many engineers don't try to compensate for very high bandwidths, and probably not many SMPS engineers are familiar with cascode circuits.

 

[godfreyl]

Also; Q1 costs money, so don't use it unless you need it (for commercial design anyway).

 

[treez]

Because many engineers don't try to compensate for very high bandwidths

Do you not think that there's more to it than just compensating for high bandwidths?
Surely the cascode stage pushes the dreadful optocoupler pole up out of the way, and allows you to achieve a far more stable power supply?
-Surley, unless you use a cascode stage, you are always going to be worrying about the massive tolerance on your optocoupler, and whether your SMPS will go unstable after say a year of operation with the customer.

Heres the schem with the cascode

Schematic with cascoded opto-transistor
https://i46.tinypic.com/vyjyo.jpg

....it does need the minus 1V bias supply to bias the circuitry, but charge pump ICs are cheap.

Can i always be sure, in the above schematic, that the opto-transistor will never saturate?

The only way to assure stability with an opto coupler, is to have a sub 200Hz loop bandwidth, which is ridiculously slow, and you may suffer output voltage overshoot if you try and speed it up without a cascode stage.

I simply cant understand why any commercial SMPS, with annual volumes of less than 30,000, would not use a cascode stage as in the schem?

 

[FvM]

The dominant pole of an opto coupler is associated with the photo transistor, not the LED. (I'm not sure which side you mean with the unusual term "opto diode").

Usual TL431 circuits already achieve a current source characteristic a higher frequencies.

A cascode stage can partly improve the frequency characteristic of a resistor loaded poto transistor, but there's still a pole of the short circuit current transfer ratio, created by Cbe||Ccb and rbe. It can be compensated, but not avoided by a cascode stage.

 

[treez]

by "opto diode" i mean the diode thats found in the optocoupler package.

partly improve

...this seems like an understatement......even the simulation shows you how achieveing stability is far far easier when the cascode is used.

By the way, is the following PNP cascode circuit going to have the same effect as the NPN one?
https://i46.tinypic.com/1zpjih0.jpg
....i mean, will this PNP circuit push up the optocoupler pole higher in frequency just like the NPN one does?

 

[godfreyl]

By the way, is the following PNP cascode circuit going to have the same effect as the NPN one?
https://i46.tinypic.com/1zpjih0.jpg
....i mean, will this PNP circuit push up the optocoupler pole higher in frequency just like the NPN one does?

Yes, and yes.

 

[treez]

OK thanks,

By the way, on page 9 (bottom right) of the article in the first post, they say that the opto transistor does not saturate because of the biasing of the base of the NPN, Q1 (pse see fig 13 on same page) to 1.5V.
But i've simulated it and it does saturate when the smps goes into no load.
The BJT base-emitter diode effectively "shorts out" the lower resistor, R11, and then the opto transistor does saturate.

How can i stop the opto transistor from saturating?
(i need a fast no load to full load response)

 

[FvM]

The photo transistor doesn't saturate if the base voltage divider is sufficiently low.

 

[treez]

The photo transistor doesn't saturate if the base voltage divider is sufficiently low.

...do you mean low value resistors in the base voltage divider resistors?

 

[treez]

..is the PNP version more likely to saturate.

I appreciate if the base divider is low enough impedance, then it shouldnt saturate, but is avoiding saturation in the opto-transistor and the PNP essential for this cascode circuit?

 

[treez]

Because many engineers don't try to compensate for very high bandwidths, and probably not many SMPS engineers are familiar with cascode circuits

i dont think "familiarity" is needed....its so simple...

im baffled why nobody is doing this...its just a BJT....and all your opto nitemares are gone?
Is there a hidden problem with this cascoding?...i cant see one

heres the ltspice sim

 

[FvM]

There's no behavioral difference between NPN and PNP cascode, except for the feedback signal polarity. For the intended feedback operation in your design, PNP cascode is appropriate,

The effect of the cacode circuit is providing a low impedance load for the phototransistor. You can achieve the same effect by omitting the series resistor R10 and the cascode transistor in your design. Then the OP will work as a I/V converter with low input impedance.