How does LDO acheive stability?

[fanrong]

Hi , everyone :

Some LDO's datasheet said that the LDO need not the ESR of the Co for
stability , but the Co must greater than some value . Do you have any idea
about the poles and zeros and their relationship ? I mean how the LDO
get stablity .

 

[penghan]

ceramic capacitors ldo stability

A regulator is unconditionally stable (i.e., has sufficient phase margin) if the open loop gain curve rolls off at 20dB/decade (i.e., like a single pole system) before crosses 0 dB.

 

[jinnose]

ldo stability without esr

fanrong, these days many LDO parts use ceramic capacitor which has almost zero ESR. In this case, LDO stability can be achieved by making one pole dominant either at output or not. In your case since Co must be greater than a certain value, the dominant pole must be at the output and the loop will be stable if other poles and zeros are out of band.

 

[fanrong]

ldo single miller

Thanks for your help , jinnose !
But the LDO use Miller compensation to make the pole at the amp be the dominant
pole . So how about this situation , jinnose ?

Another question :
What is the shortcoming and "longcoming" by make the pole at the amp instead
of the pole at the ouput being the dominant pole ?

 

[fanrong]

Re: LDO stability

Another question :
What is the shortcoming and "longcoming" by make the pole at the amp instead
of the pole at the ouput being the dominant pole ?

 

[hung_wai_ming@hotmail.com]

LDO stability

i think jinnose is rite. Nowadays all cap claims low ESR. If we didn't use dominant pole compensation with large output cap, then it's necessary we need to add extra small resistor at the output to do pole-zero compensation. Of coz, large cap cost more on PCB. Ideally, larger cap can reduce ripples as well

 

[OpAmp]

Re: LDO stability

Hi All,

I don't agree with this:

A regulator is unconditionally stable (i.e., has sufficient phase margin) if the open loop gain curve rolls off at 20dB/decade (i.e., like a single pole system) before crosses 0 dB.

Some regulators are stable when the output capacitor is higher than a certain value and some are stable when the output cap is less than a certain value. It depends on the used structure.

If you make the output pole the dominant one,then your regulator will be stable when the output cap is higher than a certain value.

If you make the amplifier pole the dominant pole of the regulator loop, then the output cap should be less than a certain value for the regulator being stable.

which one is better? It all depends on the appliction.

I hope that it helps you.

OpAmp

 

[prajit]

Re: LDO stability

Most of the times load impedance is sufficient for stability but many times manufacturers ask for Co to make it stable under transient conditions. Therefore, it is always better to check the concerned manufacturer website before using it as most of them have an application note regarding their LDO stability.

 

[penghan]

Re: LDO stability

Hi OpAmp,

Our advices are not contradict. You talk about it by case and i talk about it by principle.

Genarally speaking, there are three poles and one zero in LDO' transfer function. One pole p1 is induced by the error amplifer. Another pole p2 is induced by the load, it is dynamic as the load current changing. The third pole p3 is induced by the pass device. Last, the zero z1 is created by the output capacitor.

p1 and p2 are always lower than 0dB cross point (in magnitude plot), p3 is always higher than 0dB cross point. Thus, zero z1 can be used to cancel one of p1/p2, so as to guarantee open loop gain curve rolls off at 20dB/decade before cross 0dB. It seems like it is a single pole system. As we now, a single pole system is stable.

Thus, the ESR of output capacitor should be proper to keep z1 below 0dB cross point.

How do you thinks about it ,guys.....

Discussion continuing, come on....

 

[hung_wai_ming@hotmail.com]

LDO stability

Hi penghan,

i don't quite understand your point. if you think p1/p2 are always lower than 0dB, this is already a single pole transfer curve and no need zero to compensate. When we add output capacitor, there's extra pole above 0dB as this cap always big, so the Bode is 2-pole system and we need zero to compensate it, unless you have a VERY LARGE cap at output to be dominate and the Bode rolls off 20dB quickly enough the p2 is not seen above 0dB, then you don't need ESR as the compensation. But in this case, the -3dB frequency is very low and your LDO will respone very slow.

 

[jinnose]

Re: LDO stability

Thanks for your help , jinnose !
But the LDO use Miller compensation to make the pole at the amp be the dominant
pole . So how about this situation , jinnose ?

Another question :
What is the shortcoming and "longcoming" by make the pole at the amp instead
of the pole at the ouput being the dominant pole ?

If this is the case, the maximum value of Co should be limited for stability. Just for reference, the simple miller compensation is rarely used for LDOs since the LDO driver size is much bigger compared to 2-stage OA (so the gate capacitance of LDO driver is).

For the second question, typically the internal compensation can use smaller Co but the transient response could be worse since it uses smaller Co. Also hard to control the stability. Think about external compensation cap (more degree of freedom) versus internal compensation cap (less flexible).