Does this SEPIC LED driver have poor phase margin?

[treez]

Hello,
Does this 5W SEPIC LED driver have a poor phase margin.?
The LED current oscillates at ~500Hz (loop crossover frequency) during start-up, as it rises upwards, -however, there is absolutely no LED current overshoot, and the LED current never oscillates when it gets near its regulation value of 50mA.

The LED driver spec is:
Vin = 12V
Vout = ~97V
I(LED) = 50mA
Constant off time
Fsw = 88KHz when in regulation (50mA)

Attached is the schematic ,LTspice simulation, and waveform of LED current and output voltage at start-up

 

[SIDDHARTHA HAZRA]

You can do a impulse response stability test to check the stability of any system. It is the best way to check stability.

After the LED current settles at 50mA you can pump a high current for a very small duration (since ideal impulse can not be generated). This will pull the current up and leave it to settle. Observe the settling behavior, ripples and peak if any. If the system is stable the the current will settle back to 50mA or else it will keep oscillating.
What is the current value or what duration that you have to calculate from the cap information at that node or just do some trial and error till you get a distinct observation.

I dont think just by observing the start up profile we can comment on the stability. Stability is well measure if you disturb a settled system and check whether it is settling back or not.

Hope it will help.

 

[mtwieg]

I agree with the above post. As the converter starts up its behavior is highly nonlinear and therefore can't really demonstrate the phase margin. Allowing it to settle then applying a small perturbation to the load or line would be a better way of checking stability margins. But such a test is still only quantitative, not quantitative.

 

[SIDDHARTHA HAZRA]

I hope you meant "qualitative and not quantitative" .... I think this test is as much quantitative as other tests like bode plot or nyquist or root locus if you can model your system as a 2nd order system ....
I don't know the related equations for impulse response but instead of impulse you can also give a step current / voltage. Now for step response you will get lots of equations to check out its damping factor and phase margin by using the settling time, peak values and other information from the transient response itself. You can take help from any good book for control system discussing transient response. Here I understand that you have to approximate your system as 2nd order. But it should give you good idea.

But still if you need to analyze using conventional methods using plots ... then you have to derive the transfer function .... identify loops that are controlling your output and stabilize each of them .... in complex systems like DCDCs we do derive approximate transfer functions .... Hope this will help or just let me know if you have a better solution ....