Post Go back to editing

AD829 stability

Hello,

Is the AD829 stable in the following conditions ? What is the Phase Margin ?

- CL=150pF capacitor load

- RL=~1kohms resistor load

- Vs = +/- 5V

- Ccomp = 15pF to inverted input

I attached the schematic (VEE = -5V and VCC = +5V).

Can I simulate the phase margin with PSPICE and how ? Is the model realistic in theses conditions ?

Thanks for your answer,

Mathieu

  • Hi mbart,

    I apologize for the delay in response. I've forwarded your query to the engineer who worked on the SPICE model for AD829 regarding simulations to get the phase margin for your circuit. However, I had a spare board here and I tried out your circuit using just one of your branches as input (tried the one with the lowest gain). With a 15pF compensation cap, there is oscillation at the output (see below).

    I switched the compensation cap to 68pF (which is recommended in the datasheet for a gain of 1), and I lost the oscillation. On bench, even with the resistors set to have a gain of 0.5, the output is still stable (see below).

    If you can still make amends to your circuit, I would recommend that you go instead with a compensation cap of 68pF. The only other option would be to increase your gain to get rid of the oscillation.

    I hope this helps!

    All the best,

    Kris

  • Thanks a lot for your answer.

    I will do the same test to try to reproduce this oscillation.

    Could you tell me how did you choose 68pF Ccomp capacitor ? How could I choose the "good" capacitor value ?

    When connecting the Ccomp to the inverting input, the datasheet doesn't mention the capacitor to choose. There is some figures where Ccomp = 3pF/ 4pF or 3pF with inverting gains of -1V/V to -4V/V (pages 14 and 15).

    Ccomp = 68pF at the gain of +1V/V is mentionned when Ccomp is connecting to ground (Table 4 page 13).

    Do you have some news about stability simulation ? I hope I can use the model to choose the "good" capacitor value for stability, or that you will help me. Because when connecting Ccomp to inverting input, the capacitor value vs gain is not mentioned in the datasheet, contrary to the connection of Ccomp to ground (Table 4 page 13).

  • Hi Mathieu,

    I apologize, in my experiments, I was connecting the compensation capacitor to ground instead of at the inverting input as in your circuit. I just tried out your circuit connections on bench, and I am seeing oscillation at the output. I just heard back from the engineer, and unfortunately, the AD829 is a very old part and its SPICE model cannot be relied on to model stability properly. To be more flexible with your compensation options, would you consider using the shunt configuration? Increasing your gain would also work.

    Could you tell me more about your application? I assume that you are trying to get more slew rate out of your circuit since you are using the current feedback configuration for compensation. If it's not in your list of options to use shunt compensation or to increase your gain, maybe you could provide the requirements for your amplifier so that I could try to find an amplifier that will work for your application.

    Regards,

    Kris

  • Hi Kris,

    Thanks for you answer and I apologize for my approximative english.

    I can't change :

    - the gain of amplifier

    - the amplifier (only the AD829)

    - the type of compensation, because i need slew rate (above 60V/µs as a minimum)

    The only thing I can change is :

    - first of all, the value of Ccomp capacitor,

    - and if it is still not stable :

         - the value of capacitor load C61

         - or the value of resistors but without changing the gain

    Are you sure you tested the circuit with a capacitor load C61 = 150pF ?

    Could you please provide a screenshot of the oscillation ?

    Regards,

    Mathieu

  • Hi Mathieu,

    maybe you can try adding a small resistor at the output to snub the capacitive load.

    This should help stabilizing the part.

    Charly

  • Hi Charly,

    Yes, I know and I checked it by simulation, but I can't add a serial resistor.

    But even if I do this modification, how can I be sure that it is stable if I can't simulate the phase margin of the circuit because the model is not representative ?

    Probing the output is sufficient to prove that the circuit is stable.

    If I can't know stability by simulation, what other test can we do ?

  • Can you tie at your output an RC to ground? (R=50, and C=200pF) to see if that eliminate the oscillation?

    If that is not an option as well, then you need to reduce your CL until your amplifier is stable.

    Charly

  • 1 - I checked on my board, there is no oscillation with the the schematics attached on the first message. Could you please confirm that you measured an oscillation and show the print screen of the oscilloscope ?

    2 - If there is no oscillation like on my board, how can I prove that the circuit is stable if the Pspice model is not representative ?

    Please remember that in my first message, I didn't think that my circuit was not stable. I had never meadured an oscillation before.

    I just wanted to know :

    - if with this type of compensation, the AD829 was capable of driving such a capacitor load (150pF)

    - and how determine the phase margin by simulation

    Regards,

    Mathieu

  • Hi Mathieu,

    I apologize for the delay on this response. As I mentioned, the AD829 is a very old part, and unfortunately, I would not trust the SPICE model to be reliable in simulation. However, I revisited my circuit and we are now getting the same results on our circuit, i.e. no oscillation. In addition, I brought the capacitive load up to 560pF without seeing any oscillation at the output. One of the other methods I tried to get an idea of the stability of the circuit was to apply a step input and see how much overshoot was present at a 150pF load.

    I calculated the overshoot to be about 10%, which looks to be an acceptable figure. Hopefully, this figure will be enough to properly satisfy your requirement.

    Regards,

    Kris

  • Hi Kris,

    Great. Thank you for your answer.

    I have the same measure that in the first page but with a Ccomp = 12pF instead of 15pF.