Compensator Problem

Discussion in 'General Electronics Chat' started by bcassidy9, Jun 13, 2013.

  1. bcassidy9

    Thread Starter New Member

    Nov 27, 2012
    5
    0
    Program: LTspice
    Purpose: Work on compensation feedback networks

    I am using a simple LC circuit to work on feedback compensation. I chose the corner to be at 1000 Hz. I am using a type 3 compensator fed back into a Voltage Controlled Voltage Source. I chose to design the new crossover frequency to be 9000 Hz. Attached is the .asc file of my design. I am running into some trouble with the crossover frequency NOT being where I want it to be and there being a ton of Q introduced when the circuit decides to fall off. I am not too worried about the Q at this point; however, I do want to know why the corner is so far out? Right now it rolls off at about 90000 Hz, which is a decade above where I want it to be. If I change the multiplier from 10 to 1 on the VCVS, it comes back down to approximately 10 kHz. What am I missing here? This happens with both a type 2 and type 3 compensator. The peak seems to change with respect to the RC zero frequency of the capacitor in the LC network.

    Fundamentally I am missing something huge here. I designed the type 3 compensator correctly and have tried several designs. However, I cannot fix this problem.

    I am not looking for a solution necesarilly. If you have one, great. If not, any guidance or suggestions will be appreciated. This is my first post, so if I messed something up, sorry! haha.

    Thanks for any help!
    Brian
     
  2. LvW

    Active Member

    Jun 13, 2013
    674
    100
    Bcassidy9, if I interpret your circuit correctly, you are simulating the loop gain, correct?
    You have mentioned the cross-over frequency but NOT which gain expression you are referring to.
    More than that, it is not easy to read your circuit because of several parts which should be connected together.
    It would be easier to help you if
    * you could redraw the circuit, and
    * you would show us the Bode plot and the corresponding gain expression.

    LvW
     
  3. bcassidy9

    Thread Starter New Member

    Nov 27, 2012
    5
    0
    LvW, sorry about that. Here are a few bode plots and the schematic. I really am having trouble understanding why the crossover frequency of the entire circuit shifts an entire decade up when changing the capacitor resistance from 0.039 ohms to 1 ohm. At 0.039 ohms, the circuit crosses over at 9kHz like I wanted it to.

    However, amidst asking the question, I answered it myself. Essentially, my compensator was totally wrong. I was not looking at the gain and phase of my new transfer function I created with respect to the new crossover frequency I wanted. I was using the old compensator and wondering why it was not working. It was in fact working, but just not as quickly. It was compensating for a gain of -34dB when the cap had an ESR of 0.039, but the new cap with 1 ohm esr had a gain of -9 dB. This is substantially larger.

    In short, I was being dumb. Huge mistake. Thanks for your help.
     
  4. LvW

    Active Member

    Jun 13, 2013
    674
    100
    Hi Bcassidy9, perhaps you are satisfied now, because the result comes close to your expectation - however, there is still an error.
    You must break the loop for placing the ac source at a low-resistive node.
    And that is the opamp output. Otherwise you cannot expect to get the correct loop gain.
    May be, in your case the error is not too large. But you should try it in order to see the difference.
    I don`t know if you know Middlebrook`s general feedback theorem. (GFT). From this you can derive that you either need two injections (V and I at an arbitrary node) or one single injection (normally V) at such a low-resistive node.
    Good luck
     
Loading...