@Sophi21394 is this analysis an assignment for academic credit?
If not, what is the assignment it is associated with?

OHHHhhoooo, I want to understand things from first principle not to just for academic credit!

 

Hi Sophi,
As you are interested in PSU filtering, this PDF may be helpful.
E

 

Its DC even in DC it matter? I generally think only ac coupling polarity matter

Hi,

Some regulators still require capacitors for stability. The low dropout type are more prone to this problem.

 

Hi,

You quoted my message from post #12 but did not add a reply.

 

220µF capacitors are electrolytic, 10µF capacitors are ceramic, and 47µF capacitors are tantalum. DO correct way of thinking is - the 220µF capacitor removes larger noise, the 10µF capacitor handles the remaining high-frequency noise, the inductor reduces current noise, and the 47µF tantalum capacitor adds reliability to the circuit, especially for long-term operation ? correct or something wrong/missing ? I doubt on inductor current noise.

That sounds about right. Inductors make a large difference in the noise when coupled with output capacitors. Even small ones like 2uH can reduce noise a lot when it's higher frequency. You may even find these on the output of switching converters.

 

OHHHhhoooo, I want to understand things from first principle not to just for academic credit!

Hi,

Have you done any hard circuit analysis yet? That's like a complete analysis of a circuit.

For your circuit you can learn a lot by assuming just as few little things. First, simplify it into a single input single output filter with just one capacitor on the output and an inductor in series with that, and some resistance in series with the inductor. That forms an RLC filter circuit. It's almost like 1/2 of your circuit.

Analyze the heck out of that. Look at frequency response vs variations in R, L and C, and see what frequencies get attenuated the most. Look at the effect of load too. Do this either by direct mathematical analysis (recommended) or circuit simulator.
Try to form experiments that will answer your questions. To find out what a chance in R does, vary it for example, then repeat the analysis.
If you know how to write computer programs in at least one computer language (and you should) you can write a program that varies parameters and spits out results in tabulated form.

You can add components to the different circuit elements to make them appear more like real life circuit components too, such as series resistance for the inductors, series and parallel resistance for the capacitors, etc., etc.

Once you get that far, you can add components until you get to the original circuit, although you will probably find that you don't really need to analyze a balanced circuit once you know how the single ended one works.

The only question this won't answer is the reliability question. You'll have to look up data sheets and compare them to find out exactly why one might be more reliable than the other, and under what conditions this is true. That's if you don't believe direct web results.

I highly recommend learning some type of analysis first and foremost, because that can answer questions that nobody else on the planet might be able to answer, and you never know when this might be required for what you do. Even lumped circuit analysis can get you pretty far, after which you may want to go deeper into element size vs frequency type questions later on.
Learning a circuit simulator like LT Spice as well.

Note in all of the above, I do not know what you have studied so far so I mentioned some things you might already know about. You might take the time to mention what you have studied so far, math and circuit analysis, physics, etc.

 

Its DC even in DC it matter? I generally think only ac coupling polarity matter

I don't think they use coupled inductors for power rail filters, they use them for AC line filters. I could be wrong.

 

Hi,

Have you done any hard circuit analysis yet? That's like a complete analysis of a circuit.

For your circuit you can learn a lot by assuming just as few little things. First, simplify it into a single input single output filter with just one capacitor on the output and an inductor in series with that, and some resistance in series with the inductor. That forms an RLC filter circuit. It's almost like 1/2 of your circuit.

Analyze the heck out of that. Look at frequency response vs variations in R, L and C, and see what frequencies get attenuated the most. Look at the effect of load too. Do this either by direct mathematical analysis (recommended) or circuit simulator.
Try to form experiments that will answer your questions. To find out what a chance in R does, vary it for example, then repeat the analysis.
If you know how to write computer programs in at least one computer language (and you should) you can write a program that varies parameters and spits out results in tabulated form.

You can add components to the different circuit elements to make them appear more like real life circuit components too, such as series resistance for the inductors, series and parallel resistance for the capacitors, etc., etc.

Once you get that far, you can add components until you get to the original circuit, although you will probably find that you don't really need to analyze a balanced circuit once you know how the single ended one works.

The only question this won't answer is the reliability question. You'll have to look up data sheets and compare them to find out exactly why one might be more reliable than the other, and under what conditions this is true. That's if you don't believe direct web results.

I highly recommend learning some type of analysis first and foremost, because that can answer questions that nobody else on the planet might be able to answer, and you never know when this might be required for what you do. Even lumped circuit analysis can get you pretty far, after which you may want to go deeper into element size vs frequency type questions later on.
Learning a circuit simulator like LT Spice as well.

Note in all of the above, I do not know what you have studied so far so I mentioned some things you might already know about. You might take the time to mention what you have studied so far, math and circuit analysis, physics, etc.

I have learned the LTSpice but not any computer language. Which programming language one best suit the purpose on which software? What i think in principle we should have a noise monitoring system/ Plot in datasheet with every electrical system that will ensure us which kind of filtering we need, using such power filter is a blind guess approach to suppress noise

 

Hi,

Have you done any hard circuit analysis yet? That's like a complete analysis of a circuit.

For your circuit you can learn a lot by assuming just as few little things. First, simplify it into a single input single output filter with just one capacitor on the output and an inductor in series with that, and some resistance in series with the inductor. That forms an RLC filter circuit. It's almost like 1/2 of your circuit.

Analyze the heck out of that. Look at frequency response vs variations in R, L and C, and see what frequencies get attenuated the most. Look at the effect of load too. Do this either by direct mathematical analysis (recommended) or circuit simulator.
Try to form experiments that will answer your questions. To find out what a chance in R does, vary it for example, then repeat the analysis.
If you know how to write computer programs in at least one computer language (and you should) you can write a program that varies parameters and spits out results in tabulated form.

You can add components to the different circuit elements to make them appear more like real life circuit components too, such as series resistance for the inductors, series and parallel resistance for the capacitors, etc., etc.

Once you get that far, you can add components until you get to the original circuit, although you will probably find that you don't really need to analyze a balanced circuit once you know how the single ended one works.

The only question this won't answer is the reliability question. You'll have to look up data sheets and compare them to find out exactly why one might be more reliable than the other, and under what conditions this is true. That's if you don't believe direct web results.

I highly recommend learning some type of analysis first and foremost, because that can answer questions that nobody else on the planet might be able to answer, and you never know when this might be required for what you do. Even lumped circuit analysis can get you pretty far, after which you may want to go deeper into element size vs frequency type questions later on.
Learning a circuit simulator like LT Spice as well.

Note in all of the above, I do not know what you have studied so far so I mentioned some things you might already know about. You might take the time to mention what you have studied so far, math and circuit analysis, physics, etc.

Can you please clarify what you mean by " hard circuit analysis". What is tried till now is plot with respect to time and Ltspice default noise analysis method which give result in V/(Hz)^1/2. But that noise analysis we can only apply when circuit is but we can't use it on small circuit.
I also admit interpretation of noise is not clear to me. Like frequency response is easy to understand that is how component / system respose to different frequecy. But how to interpret this V/(HZ)^1/2. Should it be taken as analog to statistical variance (V^2)/Hz ?

 

I have learned the LTSpice but not any computer language. Which programming language one best suit the purpose on which software? What i think in principle we should have a noise monitoring system/ Plot in datasheet with every electrical system that will ensure us which kind of filtering we need, using such power filter is a blind guess approach to suppress noise

You can use almost any language even Basic. If you have not done this yet though it could take a while before you feel comfortable doing this kind of analysis.

LTSpice is pretty good for understanding the basic working of circuits. It should not be a substitute for mathematical analysis unless you don't have much math behind you. Math, along with good textbooks, tells us things about circuits we would never be able to figure out by using a simulator unless maybe we spent long hours trying to figure out key aspects of circuits through the simulations.

 

Can you please clarify what you mean by " hard circuit analysis". What is tried till now is plot with respect to time and Ltspice default noise analysis method which give result in V/(Hz)^1/2. But that noise analysis we can only apply when circuit is but we can't use it on small circuit.
I also admit interpretation of noise is not clear to me. Like frequency response is easy to understand that is how component / system respose to different frequecy. But how to interpret this V/(HZ)^1/2. Should it be taken as analog to statistical variance (V^2)/Hz ?

Yes, by Hard Circuit Analysis I meant sitting down and analyzing a circuit through mathematics and theory on circuit analysis, such as Nodal Analysis.

Understanding the noise is probably best done on a comparative basis. Look up amplifiers that are used for audio and see which ones have better noise spec's than others. Especially preamps. That might give you some insight.
Signal to noise ratio is also interesting.

Here's a link from right here on this website:
What Is Electrical Noise and Where Does It Come From? - Technical Articles