Stability issues with LDO -> BGR+LDO loop

[shyam21]

I have a LDO whose output acts as supply to a BGR and a secondary LDO. The output of the 2nd LDO acts as the reference voltage of the 1st LDO. I did stb analysis on both LDO's individually and both have around 60 deg phase margin. So i assume both are stable. Both LDO's work perfectly fine with a voltage source.

When the loop is formed, A start up circuit is used on the 1st LDO. When this start up circuit is on, the 1st LDO gets a Vref from another source and all 3 circuit work properly. When the start up circuit is turned off, the now stable 2nd LDO's output voltage is the reference voltage for the 1st LDO. During simulation, when the start up circuit is turned off, all the outputs starts oscillating at high freq. which looks like a loop stability problem. A stb analysis on the whole loop breaking the loop at the output of the 1st LDO resulted in gain less than 0 (so loop is always stable).

Any idea where the stability issues arrive from?

I tried another opamp (same LDO phase margin) but this time when the startup is turned off, all voltage goes to 0. I assumed that this is due to the slew rate of the opamp wasnt enough but even with better SR, the outputs was the same.

How can i proceed?

 

[SIDDHARTHA HAZRA]

Hi shyam,

If I clearly understand, You have two LDOs and one BGR. The first LDO acts a supply to both the 2nd LDO & BGR. After startup, when all of them have achieved steady state then you are using the output of the 2nd LDO as reference of the 1st LDO. Now if I guess correctly then the BGR is supplying reference to the 2nd LDO.

1) My question is why to use the LDO2 output as reference and make another loop when you have a BGR.
2) Even if the over all loop gain of this new third loop is zero, and if i visualize the entire scenario with two simple LDOs then the over all phase shift in the third loop is 360deg or 0deg.
-- First of all we cannot comment on PM with gain less than 0dB as we can not calculate.
-- Is 0dB necessary .... shouldn't we have some loop gain??
-- In such a case it is difficult to judge what is happening with the 3rd loop. I would suggest please introduce a 180deg phase shift in the third loop and try.

Hope this will help ....

 

[shyam21]

Hi Siddhartha,

Your understanding is correct.

1. For functional purposes really. The LDO 2 outputs close to 1.5V which is required. The BGR is a simple BJT bgr and is much more accurate close to 1.2V. Besides, the BGR has been characterized in this process multiple times before. So not touching the BGR was preferred.

2. In the current circuit, the opamps of both LDO's negative input takes the reference voltage and the resistor divided feedback goes to positive terminal. So if i alternate the inputs in one of the LDO opamps, will i be able to get 180 deg phase shift? Or were you thinking an additional circuit. Both LDO's use PMOS pass/power FETs. So one NMOS FET should also do the trick, right?

Thank you for your help.

 

[SIDDHARTHA HAZRA]

Hi shyam,

1) To make 1.5V out of 1.2V ref from BGR you need not touch the BGR. You may use an opmap. Connecting the 1.2V to the +ive and taking a resistor divided feed back to generate 1.5V at the output of the op-amp.

2) If you alternate the inputs in one LDO then their own loops are going to be unstable. So some extra ckt would help.

 

[shyam21]

Hi Siddhartha,
1. That circuit sounds like an LDO without a pass transistor. How is this more beneficial than an LDO?
2. Right. I forgot that point. An additional stage in one of the opamps will give that phase shift. Is there any other circuit you are familiar with?

Thanks again.

 

[SIDDHARTHA HAZRA]

Hi shyam,

I mis-understood. It is not a replacement of LDO ........ the picture will explains .... I am suggesting this as all the loops are independent and can be stabilized separately ..

hope this helps ...

 

[shyam21]

Thank you Siddhartha,

I now have a couple of ideas that i could try to stabilize the loop. Will see how that goes.