I think you should also include FOM. UGB/power is interesting and will yield some good details as well. This will also help to compare different compensation schemes and their efficiency. Thanks a lot for this insight.
Hello Analog Snippets! This was a great video! Could you please point me to some material from which i can learn frequency compensation/control theory when it comes to electrical engineering.
All is very nice, but there is one thing I couldn't understand: At first, we put a large capacitor parallel to the first amplifier such that its crossover point came on top of the second amplifier's pole. Then, we put a pole and a zero parallel to the first amplifier such thatthe resulting zero came on top of the second amplifier's pole. Which one of these is the dominant pole compensation?
That is a great question. Indeed just looking at the equations we can't say if it is a pole frequency or a zero. Ofcourse you can do the maths and find it out. To understand it intuitively you need to understand that at pole frequency, magnitude starts to roll off and at zero frequency magnitude stops to roll off. Also notice that roll off is a property of capacitive impedance while resistance is a constant impedance. At pole and zero frequencies, capacitive and resistive impedances are equal. At these frequencies one type switches into other. And that is the point. If impedance switches from resistive to capacitive impedance then it is pole. And if it is capacitive to resistive then it is zero. Hope this makes sense. Thanks.
@@analogsnippets Thanks for your reply! Doing the maths, I obtain same equations. But intuitively, I understand it like this, that there is a pole when the path with series impedance (Rz+ZC) starts to shunt R1 (assume that there is a current as an input flowing into the node). So in my understanding, then the pole should be calculated by R1 = Rz + ZC. (Knowing that it's obviously wrong... )
You can prove it using simple geometry. Say amp1 and amp2 are single pole amplifiers with gain g1 and g2, and bandwidth p1 and p2 (p1 < p2). If we cascade them then gain becomes g1*g2, with first pole at p1 and second at p2. In log frequency scale, cascaded amp will cross frequency axis midway between amp1 and amp2, which amounts to geometric mean in linear frequency scale. Try it.
Great videos. Please keep making them!!
Incredible insight into RC circuit & IC design proved using math.
Rarely are people able understand this well let alone make a TH-cam video out of it.
I think you should also include FOM. UGB/power is interesting and will yield some good details as well. This will also help to compare different compensation schemes and their efficiency. Thanks a lot for this insight.
Hello Analog Snippets! This was a great video! Could you please point me to some material from which i can learn frequency compensation/control theory when it comes to electrical engineering.
All is very nice, but there is one thing I couldn't understand: At first, we put a large capacitor parallel to the first amplifier such that its crossover point came on top of the second amplifier's pole. Then, we put a pole and a zero parallel to the first amplifier such thatthe resulting zero came on top of the second amplifier's pole. Which one of these is the dominant pole compensation?
At 9:55, the pole frequency can be determined by (R1+Rz) = 1/sC and the zero frequency by Rz=1/sC. Can someone explain why please?
That is a great question. Indeed just looking at the equations we can't say if it is a pole frequency or a zero. Ofcourse you can do the maths and find it out. To understand it intuitively you need to understand that at pole frequency, magnitude starts to roll off and at zero frequency magnitude stops to roll off. Also notice that roll off is a property of capacitive impedance while resistance is a constant impedance. At pole and zero frequencies, capacitive and resistive impedances are equal. At these frequencies one type switches into other. And that is the point. If impedance switches from resistive to capacitive impedance then it is pole. And if it is capacitive to resistive then it is zero. Hope this makes sense. Thanks.
@@analogsnippets Thanks for your reply! Doing the maths, I obtain same equations. But intuitively, I understand it like this, that there is a pole when the path with series impedance (Rz+ZC) starts to shunt R1 (assume that there is a current as an input flowing into the node). So in my understanding, then the pole should be calculated by R1 = Rz + ZC. (Knowing that it's obviously wrong... )
It's resistive impedance versus capacitive impedance. So R1 and Rz in series can be considered as one resistance. So it's more like Req=Zc.
at 15:34 why w0=(w01*w02)^1/2
You can prove it using simple geometry. Say amp1 and amp2 are single pole amplifiers with gain g1 and g2, and bandwidth p1 and p2 (p1 < p2). If we cascade them then gain becomes g1*g2, with first pole at p1 and second at p2. In log frequency scale, cascaded amp will cross frequency axis midway between amp1 and amp2, which amounts to geometric mean in linear frequency scale. Try it.