how big capacitor do I need?

[Tiiu]

If the current at power supply is 1A,the noise swing of supply voltage is 1V, the freq of the noise is 2000KHz.
To make the noise swing become 100mv, how big capacitor do I need to parallel connected?

 

[siongboon]

You can try electrolytic capacitor maybe 100uF to 1000uF.
Should be able to filter & stablize the supply.
Larger capacitor connected parallel to the circuit
helps to eliminate high frequency components.
I am not sure to the exact how to compute the correct capacitor value.
Typical circuit design seldom need to have precise value.
Just experiment and try, you can learn too.

Regards,
Siong Boon
www.siongboon.com

 

[echo47]

The capacitor cannot be determined from the info you have given. You also need to know the impedance of the signal it will be filtering.

More important than capacitance value, you want a capacitor that has low impedance at the 2 MHz noise frequency. That probably rules out a big electrolytic - they are typically hundreds of milliohms at that frequency. A small cap may be the best choice - maybe a few hundred nanofarads, or a few of them in parallel, but it would really help to know the signal impedance first.

You may be interested in the impedance plots on page 2 of this WIMA capacitor data sheet. The 0.47uF part is around 40 milliohms at 2 MHz, and I confirmed that by measuring an actual part. Beware, these plots can be puzzling if you've never seen one before.
https://www.wima.com/EN/WIMA_MKS_2.pdf

Another option is to put a ferrite bead in series with wherever the noise is coming from. Then the capacitor could more easily filter the noise.

By the way, how much DC voltage is on that power supply signal? A high voltage would rule out some capacitor choices.

 

[Tiiu]

Siong Boon ,Echo47 ,Thank you

this is question I know from a person who attend a analog eng. job interview. and the info at first post is all I know.

If someone could tell me a equation to caculate the capacitor would be very helpful.

 

[echo47]

If I was the interviewer, I would consider a simple calculated answer to be wrong. I would hope the candidate asks for more information, or mentions capacitor resonance, or says a few words about parasitic inductance, or at least explains what assumptions he used when making the calculation. That would demonstrate experience with practical design issues.

Some interviewers ask questions carefully worded to test the candidate's education, experience, problem-solving ability, motivation, etc. Such questions may not have a right or wrong answer, but are intended to initiate a thoughtful discussion.