SMPS (PFC) Phase margin less than 45 degrees

[eem2am]

pfc smps

Hello,

I am referring to the LT1509 Power factor corrector IC....Datasheet...

https://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1142,C1138,P1516,D1119

if you please can see page 8, Figure 3, it is evident that the phase margin (dotted) is less then 45 degrees for frequencies below the Zero gain crossover frequency.

This is surely bad design ?

I am very grateful for any thoughts, and thankyou for previous other answers which i am reviewing with gratitude.

 

[LvW]

phase margin at greater than unity

if you please can see page 8, Figure 3, it is evident that the phase margin (dotted) is less then 45 degrees for frequencies below the Zero gain crossover frequency.
This is surely bad design ?

The relevant phase margin is determined only at the zero gain crossing point.
But in general, I agree with you. The loop gain should cross the 0-db-axis - if possible - with a slope of app. -20 dB/Dec. Therefore, the loop gain - as shown in the figure - should be decreased app. by 10 dB. Then, the margin will be better and in the vicinity of 55 deg.

 

[eem2am]

smps loop response

LvW, i appreciate your response, but i beg you to reconsider...

at frequencies below the zero gain crossover frequency, the phase margin must be above 45 degrees for stability.......

i feel convinced of it.

just imagine if the phase margin was zero below the ZGXF......then you would have gain greater than one , and additive feedback.......that is a cause for oscillation , that is , instability.

I am growing convinced that this datasheet is in error....

thankyou for hearing me out, i am grateful for any thoughts on this.

 

[FvM]

smps stability

It's the closed loop response of a typical lead-lag compensation, often used with control systems. It' stable without doubt. As a serious disadvantage, it's rather sensitive to gain variations.

To my opinion, it isn't of much use to qualify control loops according to their phase margin without considering the application, particularly actuator and load dynamic and acceptable deviation.

 

[LvW]

circuit to improve phase margin

Hi eem2am,

Quote:
at frequencies below the zero gain crossover frequency, the phase margin must be above 45 degrees for stability.......
i feel convinced of it.

No, certainly not. Why above 45 deg. ?

just imagine if the phase margin was zero below the ZGXF......then you would have gain greater than one , and additive feedback.......that is a cause for oscillation , that is , instability.

No, also in this "critical" case, there would be no instability. Perhaps it is surprising for you, but there are points which fulfill the Barkhausen criterion - and, nevertheless there are no oscillations.
Barkhausen is only a necessary oscillation condition, but not a sufficient one. (This very often is forgotten or even not mentioned in most textbooks !).
Because of the nature of the transfer function under discussion (with a zero to enhance the phase) you have to apply the complete NYQUIST criterion to verify stability - and this criterion indicates stabilty !!

I am growing convinced that this datasheet is in error....

Because of the above arguments: I don´t think so.

 

[FvM]

marginless

Apart from a strict stability criterion, that has been discussed by LvW in detail, it's true that a below 45 degree phase margin possibly results in bad regulation action. It's also true, that the phase characteristic below unity gain has some influence on the closed loop behaviour, if you prefer bode plots for closed loop analysis. But it's somewhat fruitless to discuss this without considering the requirements of a particular control application.

 

[eem2am]

pfc stability loop

Hello,

FvM & LvW,

I appreciate your comments, though my learning on this is from the feedback control chapters of Marty Brown's book "Power supply cookbook"

It clearly warns against phase less than 45 degrees for trequencies less then the ZGXF.

It shows a graph showing it and saying "excess phase" and warns against it.

it is surely a fact that if the gain is greater than one and you feedback is becoming additive at the input, then youre output will spiral away...instability

 

[LvW]

gain and phase margin smps

It clearly warns against phase less than 45 degrees for trequencies less then the ZGXF.
It shows a graph showing it and saying "excess phase" and warns against it.
it is surely a fact that if the gain is greater than one and you feedback is becoming additive at the input, then youre output will spiral away...instability

Sorry, but your last statement certainly is wrong !
As I have mentioned already, in some cases the complete Nyquist stability criterion has to be applied since the simplified version which uses only the phase margin leads to errors.
I attach two documents which show you some examples for a stable system -- even when there is a frequency for which the loop gain is above 1 and the phase 360 deg.

I am sorry, but I have the book mentioned by you not available, so I cannot re-check the formulation. May be the author wants to say that perhaps the step response or something else is not ideal if the phase comes to close to the limit (for frequencies below cross-over). But, certainly, there will be no danger that the circuit becomes unstable !

 

[LvW]

smps instability

Here comes the other document.

 

[FvM]

no phase margin

Cause I was tired of the wild guesses about the output spiraling away, I simulated the load step response of a simple equivalent circuit. You can see from the bode plot that the circuit characteristic is very similar to the Linear application. There are some minor differences, as I ignored excessive phase from modulator dead time. The load step response is even better than I guessed, it's nearly aperiodical.

 

[eem2am]

unity of smps

thankyou very much LvW and FvM, i am reading this now.