Bose plot stability analsis is irrelevant for hysteretic mode SMPS?

[treez]

Hello,
Can you confirm that a Bode plot stability analysis for a switched mode LED driver based on the ZXLD1360 IC (schematic as per page 2 of the datasheet) is pointless?

That is, ZXLD1360 based LED drivers are intrinsically unstable, its just that the instability is bounded by the comparators?
So can you confirm that the phase and gain margins (if produced) would indicate instability, and so would be pointless?

**broken link removed**

 

[schmitt trigger]

Hello,
Can you confirm that a Bode plot stability analysis for a switched mode LED driver based on the ZXLD1360 IC (schematic as per page 2 of the datasheet) is pointless?

That is, ZXLD1360 based LED drivers are intrinsically unstable, its just that the instability is bounded by the comparators?
So can you confirm that the phase and gain margins (if produced) would indicate instability, and so would be pointless?

**broken link removed**

Without going into too much detail over the datasheet, yes indeed, this appears to be a hysteretic converter.
You are correct, a bode plot is irrelevant.

 

[treez]

-Thanks, then why did a gentleman with a PhD in SMPS's do a bode plot for it when I worked at a large lighting company?
Do you agree that the phase margin would be zero degrees for such a hysteretic converter

 

[schmitt trigger]

-Thanks, then why did a gentleman with a PhD in SMPS's do a bode plot for it when I worked at a large lighting company?
Do you agree that the phase margin would be zero degrees for such a hysteretic converter

Something that I do know -but just barely- about hysteretic converters is that they sometimes experience chaotic behavior.
I'm not educated enough to perform a mathematical analysis such a behavior, but my feeling is that a linear analysis tool like a Bode plot would not be the best tool for this type of converter.

 

[dick_freebird]

Because they never operate in linear small signal fashion,
linear small signal analysis can't get a grip. Maybe if you
created a state-space-average-model abstraction you
could get some idea. But I'd settle (heh) for a transient
load-step, line-step, inject-perturbation series of analyses
and just eyeball it for whether it remains within regulation
band, or doesn't.

It is basically a bistable machine, not a stable loop per se.
So you shouldn't expect classical stability even if the tools
with their built-in assumptions / abstractions / limitations
were applicable.

Many end markets (or the plaintiff's bar) will require a closed
form mathematical analysis and PhD Gentlemen love to do that
kind of thing by the billable hour. Doesn't mean it's relevant or
right; somethimes you just have to say "Yes, Sir!" and take the
paycheck.

 

[schmitt trigger]

Many end markets (or the plaintiff's bar) will require a closed
form mathematical analysis and PhD Gentlemen love to do that
kind of thing by the billable hour. Doesn't mean it's relevant or
right; somethimes you just have to say "Yes, Sir!" and take the
paycheck.

Sometimes ignorant customers require that too.
It may not make sense, but if their rule book says that one must do X, Y and Z tests, by golly one must do them, regardless whether they make sense or not.

The worst -and I'm not making this up- is an engineer who required us to supply ESD data on a #18AWG wire harness, because their manual said that every component in that particular assembly required an ESD file.

 

[mtwieg]

A hysterestic converter still has a closed-loop frequency response (probably close to a linear delay for small signal conditions), and you can certainly put that on a bode plot.

 

[schmitt trigger]

mtwieg; you are absolutely correct, the frequency/phase response vs frequency may be plotted.

However, the original poster was asking whether a Bode plot stability analysis is meaningful in an hysteretic converter.

But as dick_f so eloquently stated, the loop in a hysteresis converter is not stable but astable.

Therefore creating a Bode plot for the purposes of stability analysis is not a meaningful tool for this case..

 

[mtwieg]

Okay I see your point. A while ago someone named Carl posted on this board about their work on hysteretic converters and demonstrated how the hysteresis may be modeled as a delay (which is dependent on the hysteresis gap, and the slopes of the ramp waveforms, and therefore also the biasing conditions), thus allowing a bode plot to be derived for the open loop frequency response. That could then be used to predict the closed loop response (which would also show a singularity at the correct frequency of oscillation).

Here is the thread, and here is a paper Carl wrote explaining it. It's interesting stuff, but I haven't seen this method of design/analysis used anywhere else.

 

[schmitt trigger]

mtwieg:

thanks for the article, I learned something new