SMPS feedback divider with capacitor across lower R

feedback divider

Hello…..

I am writing about the effect on the loop gain of a capacitor which is placed across the lower feedback divider resistor of an SMPS. (Either isolated or non-isolated SMPS).

I worked in one company who made SMPS’s and they virtually always placed a parallel capacitor across the lower feedback divider resistor in order to get rid of noise.

However, one day an engineer , without telling anybody, decided to change the value of this capacitor……I am afraid I could not glean what had been the exact effect of this , but it was well known that it caused a big problem.

Apparently, the value of capacitor that he chose had an effect on the feedback loop and caused instability.

..The thing is, if you examine the transfer functions for error amplifiers, (eg Type I , Type II , Type III), they do not include this capacitor, so….

(1) ..how can we tell what the effect that this capacitor will have on the loop gain will be?

Basso’s book (Pg 256) on Switch mode power supplies has this to say about the lower feedback divider resistor……………

[R(lower) = the lower feedback divider resistor]

- - - - - - - - - - - - - - - - - -- - - - - - - - -- - - - - - - - -- - - - - - - - -
“R(lower) does not play a role in the ac response as long as the op-amp (ie, the error amp) ensures a virtual ground. Why? Simply because the op-amp maintains 0V on the inverting pin, thus making R(lower) useless for the ac analysis. However, R(lower) helps to select the needed dc output voltage, together with the upper feedback divider resistor”
- - - - - - - - -- - - - - - - - -- - - - - - - - -- - - - - - - - -- - - - - - - - -

…….So from Basso, ……
(2) …would I be correct to conclude that placing a capacitor, or indeed any component, across the lower feedback divider resistor, would have no effect on the error amplifier’s transfer function?

……However, on page 339, Basso states………

- - - - - - - - - - - - - - - - - -- - - - - - - - -- - - - - - - - -- - - - - - - - -
“…..it is important to note that the (loop) gain curve remains unaffected by the divider network as long as we have a virtual ground. This is no longer the case for a loop closed with a transconductance amplifier (OTA) as in PFC stages, for instance (MC33262).”
- - - - - - - - - - - - - - - - - -- - - - - - - - -- - - - - - - - -- - - - - - - - -

…..The UC3845 (one of the cheapest and most popular PWM chips in the world) has a transconductance amplifier for its error amplifier, so …..

(3) …would I be right in presuming that the lower feedback divider resistor does have an effect in the UC3845 error amplifier’s transfer function?

resistor error in divider

doing a filter in the feedback to reduce the noise will affect anyway the control system. The filter changes the phase and amplitude in the bode plots of the loop gain.