This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Noise on input of switching regulator, How does it affect output?

Other Parts Discussed in Thread: LM22678-Q1, LM22678


I've started a power design using TI's WEBBENCH.  The resulting design utilizes a LM22678-Q1 (-5V).  The output plots show under 3mV of ripple, which is acceptable.  However, I don't know if this is assuming 0V ripple on the input supply.  The data sheet doesn't have any PSRR specs on it, so I'm at a loss on what to expect if there is any noise on the input supply voltage.




  • Hi Ben,

    You didn't specify your Vin and Vout specs for your application. We used Webench on the LM22678 for 12V input and 5 V output the the default simulation mentions 20mv p-p. We will use this as the starting point for discussion.  In practice, the ripple will contain high frequency spikes that extend well past a 100Mhz. These high frequency spikes can be suppressed with low value ceramic bypass capacitors. Most datasheet specs for switch mode power supplies ignore the high frequency spikes and only quote p-p ripple at the switching frequency. You are also correct in observing very few datasheets for SMPS IC's, give PSRR in the same manner as is done with linear regulators. The reason is that it is difficult to properly inject the input side AC signal and get a proper measurement. In contrast this is very easy to do with Linear parts.

    Input noise that is within the bandwidth of the switcher (up to 1/2 the switching frequency) can reappear on the output of the switcher. The degree of which this is a present depends on the control topology of the IC. 



  • Hello Anston,

    thanks for responding. I've attached the design from WebBench.

    I know that my input supply will have roughly 250 mV of noise on it, and altough I haven't taken a spectral measurement, it is probably mostly in the low MHz range that I'm concerned about since that is where all my IC's PSRR and CMRR rolls off.  I do have bypass capacitors mostly in the 0.1 uF range to handle the higher frequency (I think the rolloff starts around 1.5MHz) stuff right at the ICs.

    So, with that in mind, hopefully I have a good start.  I was trying to filter as much as possible right at the input of the supply.  I have seen different methods that always seem to perform better on paper than in practice, so I've searched for guidance and found the attached paper.  I planned on implementing the Parallel Damped filter in the whitepaper.  I just wasn't sure if I needed to include the capacitors from the WebBench design (Cin and Cinx) in the input filter design.  For example: Should my final solution have my input filter components followed by Cin and Cinx, or should my input filter feed the Converter directly without Cin and Cinx?

    The whitepaper starts with an example simple switcher circuit on page 11, but then ignores Cin for the rest of the calculations.  So, am I correct to assume that the input filter replaces the Cin found in all basic switcher designs?

    Thank you for your assistance.



  • Ben,

    Your input filter should be able to replace the bulk input cap or the Cin cap from the Webench schematic. But I would still suggest having the smaller Cinx right between the input and ground pins of the IC. The path between the Cinx ground and the anode of the diode should be kept very short on the layout. I hope this helps.


  • Akshay,

    Thank you.