• State Suggested Answer
  • Replies 2 replies
  • Answers 1 answer
  • Subscribers 42 subscribers
  • Views 32 views
  • Users 0 members are here
Options
Related

Reducing the feedback loop bandwidth of the ADP7112 seems not to work in LTSpice

Bast!
on Jan 21, 2021

Hello,

I have a noisy circuit that I needs to be isolated from the main supply. Space constraints make a filter in the kilohertz range not suitable for our design.

The noisy part can easily live with some 100 mVpp of ripple or oscillations on its own supply, though. So the idea is to use a LDO and simply reduce its feedback loop bandwidth.

This does work well in several simulations with ICs from different vendors.

Unfortunately it does not seem to work in LTSPice with the ADP7112 (which is preferred because of its small size in a WLSCP package). Don't know if this the ICs real behavior or just an artifact in the simulation or if the feedback loop bandwidth is nor really part of the model.  If I change the feedback loop bandwidth by adjusting the capacitor at the adj-pin it seems to little no effect. Works well on the ADP163 though.

Simulation file attached

ADP7112 -reduce feedback loop bandwidth .asc

Best regards

Basti

  • 0
    •  Analog Employees 
    on Jan 22, 2021 8:46 AM 1 month ago

    Hi Basti,

    That waveforms you're seeing is just because of how the ADP7112 was modelled. Ripple rejection going to the input is present on the actual device but may not be as accurately modelled in the LTspice. 

    I highly suggest primarily increasing the COUT to improve ripple filtering on both directions.

    Regards,

    Bryan

  • 0 on Jan 22, 2021 8:59 AM 1 month ago in reply to bgonzales

    Hello Brian,

    I implement more/larger capacitors if I had the space to do so. If I had space the way to go would be a L-C low pass filter with a cutoff roughly at 10 kHz and a current capability of 100 mA (or simply more capacitance to suppress the noise).

    So what I am doing is using the LDO with a highly reduced feedback loop bandwidth (and therefore PSRR) to isolate the input rail from the output.  Normally it is the other way...

    Looks like I have to get the evalboard to run some real world tests.

    Best regards

    Basti