Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Inverse engineering : from a transfer function to an electronic circuit systems !

Status
Not open for further replies.

hananeel

Newbie level 3
Joined
Mar 25, 2008
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,307
Design control systems in matlab and other software today its very easy, get for example the transfer function of a circuit , then optimize it,... its kinda easy these days,...

what I'm looking for,... its a easy way , once I have my electronic system, mathematically speaking - get the design of the real circuit with its devices like loads, capacitors, diode, transistor, etc etc ,...

so, I can Begin from the equations/ and my system responses, and then develop the actually electronic system,..

well,...
thanks for any idea,..
 

Yeah, you can. That's something they show you in sophomore/junior EE classes (s domain).

Take a look at circuit equations for RLC circuits when taking their Laplace transforms. It will give you elements like R, Ls and 1/sC. This should like vaguely familiar to transfer functions, which are functions of s, i.e. H(s). When you substitute s = jω, you can plot the response of the circuit versus frequency (Bode plots), which is handy in designing/analyzing filters. That filtering is what causes you to speed up or slow down control loops, depending on the type of filter/circuit you implement in the feedback path of your loop.

**broken link removed**
 

Yeah, you can. That's something they show you in sophomore/junior EE classes (s domain).

Take a look at circuit equations for RLC circuits when taking their Laplace transforms. It will give you elements like R, Ls and 1/sC. This should like vaguely familiar to transfer functions, which are functions of s, i.e. H(s). When you substitute s = jω, you can plot the response of the circuit versus frequency (Bode plots), which is handy in designing/analyzing filters. That filtering is what causes you to speed up or slow down control loops, depending on the type of filter/circuit you implement in the feedback path of your loop.

**broken link removed**


suppose I have f(s) = (s^2 + 5s^4 ) / (2s^4 + 3s^2 +9s) for example,... now,... i want to develop the electronic circuit
how you would do it ?,...

PS: thanks for the pdf link,.. !!
 

In filter design the traditional way is to use a ladder topology, as in File:Cauer lowpass.svg - Wikipedia, the free encyclopedia.

If you sit down and work out the impedence at the input, it turns out to be:
\[\frac{1}{sC_1 + \frac{1}{sL_2 + \frac{1}{sC_3 + \cdots}}}\]

So, by finding the impedence of the network that implements the transfer function (I cannot recall how exactly to do this, or I would give some more details; it involves two-port networks), and representing it as a continued fraction, you can read off the component values.

The nature of the transfer function obviously has an effect on the networks that you can use; looking at the schematic above it should be clear that the filter is by nature low-pass.

Long story short: If you want to produce an analogue implementation, grab a book on analogue filter design and you will find information on the synthesis of arbitrary transfer functions. Or failing that use a digital controller and just read off the z-coefficients to get your filter.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top