Talk:Frequency compensation

Several comments from an anonymous editor
From my understanding of frequency compensation of amplifiers subject to feedback, this article may still need some work. For example,

1. There is no mention, in the "Explanation" paragraph, that the amplifier will oscillate under certain gain conditions when the phase of the output signal reaches -180°. As I understand it, prevention of amplifier oscillation is the principal reason for frequency compensation. The relevant gain condition at -180° is open loop voltage gain in excess of the amplifier's intended closed loop gain. Without this gain, the amplifier will be at least marginally stable. Even in dominant pole compensation schemes, there is a point where the output phase is reversed but the amplifier is stable because of the lack of such excess gain at that point. However, there is no mention of the gain issue at all.

2. The fact that there are compensation schemes other than the common dominant pole method is not mentioned. One example is lead compensation, in which a zero is added to the output response to cancel one of the poles. Another is lead-lag compensation where both a zero and a pole are added to the output response.

3. It is stated in the first paragraph of the "Drawbacks" section that the bandwidth lost through the use of dominant pole compensation can be restored with negative feedback. In my understanding, this is incorrect. The closed loop gain can never exceed the open loop gain at any frequency (and the open loop gain at higher frequencies is what is compromised by this compensation). Without open loop gain in excess of the closed loop gain, feedback can have no effect.

4. It is implied in the second paragraph of the "Drawbacks" section that   the introduction of active devices in the feedback loop is a drawback of  frequency compensation. This is not always true. However, the problem is not a drawback of compensation but a failure of the compensation used to overcome the additional phase lag created by the "in loop" devices. Besides, inclusion of these devices is not always disastrous; state variable filters are commonly implemented in this way.

Does anyone disagree?Anoneditor 07:08, 1 September 2006 (UTC)

Question: The next-to-last sentence in the first paragraph of the "Explanation" section states, "If a flat frequency response over the passband is of primary importance, then frequency compensation may not be necessary or beneficial." What is this intended to mean? If the author uses the term "passband" in its usual sense (the -3db voltage point), then frequency compensation is for the most part irrelevant to passband flatness. In most cases (dominant pole), the compensation will merely determine the upper frequency limit of the passband. If the author uses "passband" to mean the amplifier's open loop gain at all frequencies, then it should be so stated. Otherwise, I fear the existing statement may confuse readers not having extensive backgrounds in this area.

Am I missing something here? Anoneditor 19:13, 1 September 2006 (UTC)

Compensator definitions are fiction here
The lead-lag compensator article (which needs to be moved to "lead–lag compensator" for WP:DASH and should be redirect linked from lead-lag compensation and lead–lag compensation) gives much better definitions of lead compensation, lag compensation, and lead–lag compensation. It is wrong to say that lead–lag is a general term describing either lead or lag, like this article currently says. A lead–lag compensator is a lead compensator followed by a lag compensator. Likewise, it is wrong to say that a lead compensator simply adds a zero. Both lead and lag compensators are minimum phase by design; they both have one zero and one pole. Otherwise they pollute the entire phase response rather than just the small section where it is needed. So someone please give a shot at rewriting those definitions (and add lag compensation). &mdash;TedPavlic (talk/contrib/@) 21:19, 21 January 2010 (UTC)

Someone please convert scans of notebook to typed formulae
Someone please convert scans of notebook to typed formulae. 65.216.171.130 (talk) 21:56, 6 August 2018 (UTC)