On Fri, 21 Jun 2013 11:34:18 -0500, Richard Owlett <rowlett@pcnetinc.com> wrote:>Tauno Voipio wrote: >> On 21.6.13 4:43 , Richard Owlett wrote: >>> In a sense this is a "homework" question. >>> The catch is it is homework from 50 years ago in a first >>> semester of DC >>> circuits ;/ >>> >>> The immediate problem is quite simple. It is a 2 port network of 6 >>> components with 2 internal nodes. It is driven by a ideal >>> voltage source >>> and I want to solve for the open circuit output voltage. It >>> actually >>> would have been suitable homework. >>> >>> Pointer to suitable refresher site please. >>> P.S. I do expect to say *DUH* when recalling how to solve. >> >> >> The canonical way is to write the loop equations, using Ohm's and >> Kirchhoff's rules: >> - the voltage over a resistor is resistance multiplied by current, >> - the sum of currents in any node is zero, >> - the sum of voltages around a loop is zero. >> >> There will be a linear equation group to solve, but it may be >> tedious. > >Chuckle. What I need is the refresher in writing down those >linear equations. > >I can write lots of valid equations. I've got myself trapped in >an infinite loop mentally where the number of equations <> number >of unknowns ;) That's why I asked for a pointer to instructional >material.Go to YouTube and search on Nodal Analysis or Mesh Analysis and see which suits you better. There are multiple instructional vids for each, and they're aimed at new students, so they should be reasonably straightforward (I haven't scoped them out for you). ;) Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
A "back to basics" question
Started by ●June 21, 2013
Reply by ●June 21, 20132013-06-21
Reply by ●June 21, 20132013-06-21
In article <OsydnZQZG-RkGFnMnZ2dnUVZ5vqdnZ2d@giganews.com>, Tim Wescott <tim@seemywebsite.com> wrote:>Now, solve the resulting linear algebra problem (I use Maxima if I want >to do this symbolically, Scilab if numerically. Masochism is >instructive, but one must have limits).It's gems like this that really validate my time (mostly lurking) on newsgroups. I'd never heard of Maxima. My need for symbolic solvers is quite rare, but when you need one, it's nice to have. Last time I did, I struggled with Googling around, but never found what I wanted. This looks to fit the bill nicely. It's only a 31 year old program.... Thanks Tim --Mark
Reply by ●June 21, 20132013-06-21
On Fri, 21 Jun 2013 21:07:32 +0000, Mark Curry wrote:> In article <OsydnZQZG-RkGFnMnZ2dnUVZ5vqdnZ2d@giganews.com>, Tim Wescott > <tim@seemywebsite.com> wrote: >>Now, solve the resulting linear algebra problem (I use Maxima if I want >>to do this symbolically, Scilab if numerically. Masochism is >>instructive, but one must have limits). > > It's gems like this that really validate my time (mostly lurking) on > newsgroups. I'd never heard of Maxima. My need for symbolic solvers is > quite rare, but when you need one, it's nice to have. Last time I did, > I struggled with Googling around, but never found what I wanted. This > looks to fit the bill nicely. It's only a 31 year old program.... > > Thanks TimFor casual work you almost certainly want WxMaxima. It's regular old Maxima inside of a nice wrapper. You still have something that's command- line-ish, but it's much easier to use. If you like spending $$ on programs and if you use Windows, look at MathCad and Maple. Maple is the bux-up answer to Maxima; MathCad (and I think Maple) gives you an environment like a white board that does the math for you, and I know that MathCad also lets you assign dimensions and keeps track of them for you (meaning -- you can add feet and meters, or inches and cubits, but if you try adding inches and quarts, or try taking the logarithm of a volt, it'll bitch at you). -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●June 21, 20132013-06-21
On Saturday, June 22, 2013 1:43:02 AM UTC+12, Richard Owlett wrote:> In a sense this is a "homework" question. > > The catch is it is homework from 50 years ago in a first semester > > of DC circuits ;/ > > > > The immediate problem is quite simple. It is a 2 port network of > > 6 components with 2 internal nodes. It is driven by a ideal > > voltage source and I want to solve for the open circuit output > > voltage. It actually would have been suitable homework. > > > > Pointer to suitable refresher site please. > > P.S. I do expect to say *DUH* when recalling how to solve.Whole range of methods using Ohms law, kirchoffs laws, nodal analysis,Thevin and so on. The Thevenin method can ease things a bit by in some cases avoiding solving simultaneous equations.
Reply by ●June 22, 20132013-06-22
Hi,
either write one equation per node, where the sum of currents flowing into
that node (resulting from difference between node voltages) sums up to
zero.
Alternatively, draw all possible loops in the circuit and name it a
fictional current. State for every loop that the sum of voltages (caused by
the currents across each element) sums up to zero.
The first method is easier to automate (matrix "stamps") for elements. I
usually use the 2nd method myself.
If the task is a textbook example, you can do it with pen and paper. For
real-world circuits, open-source Maxima does the linear equation solving
for me.
If anybody is interested in Maxima - which is by no means mandatory but
sometimes very useful - here are some example commands. I don't have a
complete example to post, unfortunately.
kill(all)$
/* circuit equations */
eq1: vIn = ii1 * (Rs + X1) - ii2 * X1$
eq2: 0 = - ii1 * X1 + ii2 + ... ... ...
eq3: ... ...
eqn: vOut = ...
eqSys: [eq1, eq2, eq3, ... ... ...]$
/* substitute frequency-dependent impedance */
eqSys:ev(eqSys, X1 = 1 / (%i*omega*C1))$
..
..
..
/* solve for single equation in vIn / vOut */
eqSys : eliminate(eqSys, [ii1, ii2, ii3, ... ... ...])$
/* substitute voltage gain */
eqSys : ev(eqSys, vOut = G * vIn)$
eqSys : fullratsimp(solve(eqSys, G))$
eqSys : ev(G, eqSys);
exportExpr : fullratsimp(eqSys);
stringout("C:/temp/transferFunctionExport.txt", exportExpr);
_____________________________
Posted through www.DSPRelated.com
Reply by ●June 22, 20132013-06-22
Richard Owlett wrote:> In a sense this is a "homework" question. > The catch is it is homework from 50 years ago in a first semester > of DC circuits ;/ > > The immediate problem is quite simple. It is a 2 port network of > 6 components with 2 internal nodes. It is driven by a ideal > voltage source and I want to solve for the open circuit output > voltage. It actually would have been suitable homework. > > Pointer to suitable refresher site please. > P.S. I do expect to say *DUH* when recalling how to solve. >One of the pages I was referred to triggered insight after a night's sleep. It was the visual format used that was the trigger. My underlying error was skipping steps trying to "solve by inspection" ;) Now to practice solving a set of 3 equations with 3 unknowns. Thanks to all.
Reply by ●June 22, 20132013-06-22
dszabo wrote:>> dszabo wrote: >>>> On 21.6.13 4:43 , Richard Owlett wrote: >>>>> In a sense this is a "homework" question. >>>>> The catch is it is homework from 50 years ago in a first semester of > DC >>>>> circuits ;/ >>>>> >>>>> The immediate problem is quite simple. It is a 2 port network of 6 >>>>> components with 2 internal nodes. It is driven by a ideal voltage >>> source >>>>> and I want to solve for the open circuit output voltage. It actually >>>>> would have been suitable homework. >>>>> >>>>> Pointer to suitable refresher site please. >>>>> P.S. I do expect to say *DUH* when recalling how to solve. >>>> >>>> >>>> The canonical way is to write the loop equations, using Ohm's and >>>> Kirchhoff's rules: >>>> - the voltage over a resistor is resistance multiplied by current, >>>> - the sum of currents in any node is zero, >>>> - the sum of voltages around a loop is zero. >>>> >>>> There will be a linear equation group to solve, but it may be tedious. >>>> >>>> Depending on your circuit, it may be possible to simplify it >>>> using the wye-delta transform and Thevenin's or Norton's theorems. >>>> >>>> -- >>>> >>>> Tauno Voipio >>>> >>> >>> Circuit simulators are super good at solving 'tedious' networks. >>> >> >> One should not use canned solutions when not able to solve >> problem the tedious way. I speak from decades of experience. >> >> > > A moderately complex circuit analysis may involve assumptions. I would > require a simulation or physical measurement (or both) of such a network to > validate those assumptions before commiting to component values. > > Measure twice, cut once. >I had to re-read your post (after a night's sleep) to see where you were coming from. I'm doing analysis not synthesis. I have the network topology and typical values. I need the transfer function.
Reply by ●June 22, 20132013-06-22
On Sat, 22 Jun 2013 11:57:44 -0500, Richard Owlett wrote:> dszabo wrote: >>> dszabo wrote: >>>>> On 21.6.13 4:43 , Richard Owlett wrote: >>>>>> In a sense this is a "homework" question. >>>>>> The catch is it is homework from 50 years ago in a first semester >>>>>> of >> DC >>>>>> circuits ;/ >>>>>> >>>>>> The immediate problem is quite simple. It is a 2 port network of 6 >>>>>> components with 2 internal nodes. It is driven by a ideal voltage >>>> source >>>>>> and I want to solve for the open circuit output voltage. It >>>>>> actually would have been suitable homework. >>>>>> >>>>>> Pointer to suitable refresher site please. >>>>>> P.S. I do expect to say *DUH* when recalling how to solve. >>>>> >>>>> >>>>> The canonical way is to write the loop equations, using Ohm's and >>>>> Kirchhoff's rules: >>>>> - the voltage over a resistor is resistance multiplied by >>>>> current, >>>>> - the sum of currents in any node is zero, >>>>> - the sum of voltages around a loop is zero. >>>>> >>>>> There will be a linear equation group to solve, but it may be >>>>> tedious. >>>>> >>>>> Depending on your circuit, it may be possible to simplify it using >>>>> the wye-delta transform and Thevenin's or Norton's theorems. >>>>> >>>>> -- >>>>> >>>>> Tauno Voipio >>>>> >>>>> >>>> Circuit simulators are super good at solving 'tedious' networks. >>>> >>>> >>> One should not use canned solutions when not able to solve problem the >>> tedious way. I speak from decades of experience. >>> >>> >>> >> A moderately complex circuit analysis may involve assumptions. I would >> require a simulation or physical measurement (or both) of such a >> network to validate those assumptions before commiting to component >> values. >> >> Measure twice, cut once. >> >> > I had to re-read your post (after a night's sleep) to see where you were > coming from. > > I'm doing analysis not synthesis. > I have the network topology and typical values. I need the transfer > function.Actually the next step in this is where a simulator comes in handy -- the next step being the Monte Carlo simulation where you vary all the component parameters and see if all the members of the resulting family of transfer functions behave nicely. -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
Reply by ●June 22, 20132013-06-22
Tim Wescott wrote:> On Sat, 22 Jun 2013 11:57:44 -0500, Richard Owlett wrote: > >> dszabo wrote: >>>> dszabo wrote: >>>>>> On 21.6.13 4:43 , Richard Owlett wrote: >>>>>>> In a sense this is a "homework" question. >>>>>>> The catch is it is homework from 50 years ago in a first semester >>>>>>> of >>> DC >>>>>>> circuits ;/ >>>>>>> >>>>>>> The immediate problem is quite simple. It is a 2 port network of 6 >>>>>>> components with 2 internal nodes. It is driven by a ideal voltage >>>>> source >>>>>>> and I want to solve for the open circuit output voltage. It >>>>>>> actually would have been suitable homework. >>>>>>> >>>>>>> Pointer to suitable refresher site please. >>>>>>> P.S. I do expect to say *DUH* when recalling how to solve. >>>>>> >>>>>> >>>>>> The canonical way is to write the loop equations, using Ohm's and >>>>>> Kirchhoff's rules: >>>>>> - the voltage over a resistor is resistance multiplied by >>>>>> current, >>>>>> - the sum of currents in any node is zero, >>>>>> - the sum of voltages around a loop is zero. >>>>>> >>>>>> There will be a linear equation group to solve, but it may be >>>>>> tedious. >>>>>> >>>>>> Depending on your circuit, it may be possible to simplify it using >>>>>> the wye-delta transform and Thevenin's or Norton's theorems. >>>>>> >>>>>> -- >>>>>> >>>>>> Tauno Voipio >>>>>> >>>>>> >>>>> Circuit simulators are super good at solving 'tedious' networks. >>>>> >>>>> >>>> One should not use canned solutions when not able to solve problem the >>>> tedious way. I speak from decades of experience. >>>> >>>> >>>> >>> A moderately complex circuit analysis may involve assumptions. I would >>> require a simulation or physical measurement (or both) of such a >>> network to validate those assumptions before commiting to component >>> values. >>> >>> Measure twice, cut once. >>> >>> >> I had to re-read your post (after a night's sleep) to see where you were >> coming from. >> >> I'm doing analysis not synthesis. >> I have the network topology and typical values. I need the transfer >> function. > > Actually the next step in this is where a simulator comes in handy -- the > next step being the Monte Carlo simulation where you vary all the > component parameters and see if all the members of the resulting family > of transfer functions behave nicely. >ROFL *GRIN* ! I'm doing things neither the normal nor easy way for some personal educational reasons. In fact the circuit I'm working with is rather well known - "Parallel Twin-T" filter. I'm (re)learning a awful lot as I trace long forgotten paths. I do have an eventual possible application in mind. If I get too frustrated I'll just download LTSpice and let it do its thing. I might use it as an example to check out Maxima - I always get more out of my own problems than those given in instruction manuals/textbooks.
Reply by ●June 23, 20132013-06-23
