0
\$\begingroup\$

In https://www.analog.com/media/en/technical-documentation/data-sheets/13756fd.pdf referring to this figure enter image description here

it says enter image description here

Why is this not a problem? My reasoning would be: phase gets close to 0°, so the feedback becomes positive with enough gain. I understand this would not be a problem if the gain at this point was less than 1 (0dB) and the signal fed back was therefore attenuated. But here we have a lot of gain which amplifies this fed-back portion of the signal and should create positive feedback. So why is phase getting close to 0° a problem near unity-gain rather than at this phase dip at higher gains? What does this mean for the stability criteria: only what happens around unity-gain matters?

\$\endgroup\$
5
  • \$\begingroup\$ Just a quick glance suggests at least two poles operating and then at least three zeros arrive to counter them, enough to move the phase further away from zero for a while. The margin is worrisome, since it cannot be taken as gospel. Variations in components, temperature, etc., all suggest more investigation is warranted. But if the margin holds for all those vagaries, then it may be workable. (I've not read the linked paper. Just looked at the figure.) Opinion of a hobbyist. I'll quickly defer to experts. \$\endgroup\$
    – periblepsis
    Commented Jun 26, 2023 at 9:03
  • \$\begingroup\$ Anyplace in the closed loop response where you have gain and inadequate phase margin could create stability problems, not just where the gain is unity. SMPS's I've seen usually have at least 30 deg phase margin out to unity gain. I'm not an expert in this area and will defer to the experts. \$\endgroup\$
    – SteveSh
    Commented Jun 26, 2023 at 10:42
  • 1
    \$\begingroup\$ For stability considerations, what matters is the phase margin read at crossover, when the loop gain is unity or 0 dB. You may have so-called conditional stability, e.g if the loop gain changes and crosses unity in an area where the phase margin is poor, then, you can experience instabilities. It would be the case in your figure if the loop gain drops by 40 dB, then you would have a new 0-dB crossover and should check if 30° is acceptable (usually the lowest acceptable limit). I have seen systems whose open-loop gain had the phase close to 0° where the gain was still high, no problem at all. \$\endgroup\$
    – Verbal Kint
    Commented Jun 26, 2023 at 11:52
  • \$\begingroup\$ @VerbalKint Thanks for directing me to "conditional stability", that helped me know what to look for so I could answer my own question. \$\endgroup\$
    – Hyp
    Commented Jun 29, 2023 at 16:03
  • \$\begingroup\$ @Hyp, no problem at all, you can have a look at my seminar or other documents in my webpage related to stability analysis. \$\endgroup\$
    – Verbal Kint
    Commented Jun 29, 2023 at 17:09

1 Answer 1

Reset to default
1
\$\begingroup\$

After Verbal Kint's tip-off of this being conditional stability I found the following link that explains the mechanism nicely: https://www.venableinstruments.com/venable-vault/why-stable-systems-do-not-oscillate

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.