Buck converter with external error amp and negative feedback?

[treez]

Hello,

I have implemeneted an external error amp with this 12V, 10A buck converter with an op amp (U3) followed by an NPN (Q1).

The error amplifier feedback is taken into the "non-inverting input" of the op amp due to the fact that the NPN (Q1) does an inversion.

Is this OK?......i always thought feedback had to go to the inverting input of the error amp?
(the error amp is external as this buck will be paralleled with ten other bucks all feeding the same 100A load...just one error amp feeds all controllers)

Schematic (also provided in pdf)
https://i39.tinypic.com/317ccx5.jpg


LTspice sim also attached

 

[treez]

it is a new design i came up with.

The op amp is inverting , even though the feedback comes back to its (+) input.......its inverting because of the NPN inverter in the feedback loop, which negates the positive feedback, making it "negative" feedback, but i wondered if this is ok?......................i mean, the TL431 does it like this, so i presume its ok?

Page 9 (top) of tl431 datasheet.
https://www.ti.com/lit/ds/symlink/tl431.pdf

 

[FvM]

i always thought feedback had to go to the inverting input of the error amp?

There are many examples demonstrating the contrary. You have to determine the overall loop gain.

As you can easily see, the error amp implements negative feedback. Stability must be determined seperately and can be a problem with gain blocks inserted in the feedback loop. It seems to be O.K. in the present circuit.

I wonder however why the intended frequency characteristic can't be implemented with the internal LT1243 error amp?

 

[treez]

I wonder however why the intended frequency characteristic can't be implemented with the internal LT1243 error amp?

...this is an excellent point, and the fact is i havent told you the whole story.........this converter will be used with ten others on parallel, (interleaved). There will only be one error amplifier, feeding all the 10 pwm controllers, whose internal error amps will be bypassed.

As you know, I cant let each controller have its own error amp as one may hog more than its share of the output current.

 

[D.A.(Tony)Stewart]

The question begs for more detail because as it stands, there are many issues overlooked. A multiphase driver is OK with a common feedback, but the implied delay from a TDM MUX multi-phase driver in feedback may contribute to loss of phase margin. The overall design needs a phase margin analysis as FvM indicates with each block transfer function well defined. Simply trying to use negative feedback willy-nilly without attention to parasitic phase shift in the external feedback circuit will lead to instability, which can be measured by phase margin or overshoot ringing or in this case more likely steady oscillation like a class D amplifier.

 

[treez]

thanks but if one of the individual bucks is stable, then the whole ten-in-parallel buck arrangement will be stable. this is because the effective switching frequency is ten times each ondividual, and this ends up meaning that the gain and phase margin become excellent.

The feedback signal does have to get shipped back to each controller on a dedicated pair of pcb traces for noise care...but thats not too bad