TPS61021A: loop stability analysis

Hi Guillaume,

Good to see you again and thanks for your questions. I see not much difference between your method and mine. For the test results with only 1uF capa on output, the gain margin is low. This could be caused by two reasons:

1.1uF capacitor capacitance is too low at that DC bias. Or 1uF is not sufficient to keep gain margin enough. You need to add or increase the capacitance;

2.Input filter. Refer to below application note. Without well design, input filter has something to do with gain margin at high frequency. You need to add a big capacitor at input side to check if the loop is influenced by input network. 

https://www.ti.com/lit/pdf/snva801

-Wenhao

Hello Wenhao,

Following your advice we made a few more measurements.
However none of the 2 options seems to resolve the issue because the peak at 300kHz is always there.

1. Measurement with 4.7uF output capacitor and 2.5V input voltage :


2. Measurement with 10uF added at the input and 2V input voltage :


There seems to be some kind of resonance in the circuit, perhaps between the inductance and the output capacitance.
What are your thoughts about that ?

Thank you
Guillaume

Hi Guillaume,

I am in Holiday until 10.10. Will reply you the week after that. Thanks!

-Wenhao

Hi Guillaume,

Sorry for delay. I think you should keep the device in PWM so its bode plot is of most meaningful. TPS61021A would enter power save mode in light load. Let me check with design of threshold entering PSM.

TPS61021A implements valley current COT control and there should be no resonance in gain plot, as inductor has been "cancelled" by valley current sensing.

-Wenhao

Hello Wenhao,

Thanks for your reply.
However the reason we are doing these stability analysis is to make sure the TPS61021A will remain stable during startup with a slew-rate controller (ref here https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1024194/tps61021a-slew-rate-limiter/3806487).
Startup duration is around 1ms and load is always light at that point (we wait a little before enabling the load). So during startup the TPS61021A  mainly charges the 10uF capacitor.

I think the TPS61021A operates in PFM during this phase.
Measurements during startup ( yellow is slew rate output, green is TPS61021A output) :

That's why we made the analysis with light loads for the 1uF only case.

Thanks
Guillaume

Great that really helps. Very good to know that PSM is always stable.

I have two comments :

Stability :
You said "The stability in PWM needs to be checked with only 1uF caps".
But if we make sure TPS61021A is in PSM during startup (TPS61021A softstart and slew rate controller activation) then we always have 1uF+10uF in PWM mode so no need to check PWM stability with only 1uF. Do you agree ?

Soft-start :
Charge current of the 1uF capacitor is much higher than 20mA at some point during TPS61021A softstart. I didn't measure it but given the input voltage drop (100mV on 2 AA), I guess a few hundred mA at the output. Perhaps we need to make sure we are effectively in PSM during all TPS61021A soft start ?

Guillaume

Hi Guillaume,

Could you please see my reply below?

1.But if we make sure TPS61021A is in PSM during startup (TPS61021A softstart and slew rate controller activation) then we always have 1uF+10uF in PWM mode so no need to check PWM stability with only 1uF. Do you agree ?

[WW]Yes and no. If 1uF only exists in PSM, then it is not necessary to check stability;

Charge current of the 1uF capacitor is much higher than 20mA at some point during TPS61021A softstart. I didn't measure it but given the input voltage drop (100mV on 2 AA), I guess a few hundred mA at the output. Perhaps we need to make sure we are effectively in PSM during all TPS61021A soft start ?

[WW]You could measure the inductor current. If its valley point stays above 200mA at start-up after Vout>Vin, then it means the device is in PWM when SS. If inductor  valley point is below 200mA, then it means device is in PSM when SS.

-Wenhao

Ok Wenhao, thank you.