I've been re-learning long forgotten circuit theory/ analysis to aid in my understanding of audio design. Having been out of college for approximately 10 years, I've forgotten more than I remember! Plus, transposing is something that I am not all that great at! oh well, I guess I will keep up practice if it is a usable skill. The struggle is real! lol
I've been using it for over 30 years. Most recently, earlier this monthI don't ever see a time when simulators replace its utility.
Maybe I misunderstood, but doesn't spice use the node voltage method to do its calculations?
I should have paid more attention in school! I wasn't focused on learning back then.
Last edited:
Yes, it does ... for specific element values.
Does it offer nodal analysis for general (not numerically defined) components (R1, C3, L2, etc) yet? I've been out of the loop for a long time
Does it offer nodal analysis for general (not numerically defined) components (R1, C3, L2, etc) yet? I've been out of the loop for a long time
That is a question that I can't answer! I'm in the very early stages of learning LTSpice IV and PSUD. This whole learning adventure started when I started building a DIYA AB amplifier and realized how little I actually know!
To be a little pedantic, I believe Spice uses Modified nodal analysis.
Otherwise there are "divide by zero" type problems with sources, a potential source has zero impedance.
Like you I became curious about the theory behind the simulation, I would love to learn exactly how LTSpice handles this, if anyone knows more.
Best wishes
David
Yes, it does ... for specific element values.
Does it offer nodal analysis for general (not numerically defined) components (R1, C3, L2, etc) yet? I've been out of the loop for a long time
No simulator I know of can output a symbolic equation. Many can handle parameters and expressions with parameters, but you have to give all parameters a value before running the simulation and you get numerical values as output.
By the way, when Alan Turing wrote down his system design for the ACE stored-program digital computer in late 1945, circuit simulation was already one of the applications he had in mind!
Last edited:
Thanks! That was my understanding (and memory), as well.
So, when I want to derive the frequency domain ("s" domain) input/output relationship of an RLC circuit, for example ... with non-numerical symbols (like R1, L3, C2) instead of specific values (1 kohm, 3 mH, 5 uF) ... i still do it by hand
pen and paper baby !!!
This is beyond my skill level. Every time I ask a question on this forum, I realize that I know less than previously thought. Eventually, I might realize that I know nothing!
Maple, and MATLAB with an added toolbox, can do symbolic math.
https://www.maplesoft.com/products/maple/features/symbolicnumericmath.aspx
Symbolic Math Toolbox - MATLAB
are you guys trying to take away my pen and paper?
Seriously ... thanks for the links! Worth looking into
The early SPICE papers are available online
here and
here.
They give some insight into how SPICE does it. However KCL and KVL (node analysis) are the key to ANY electrical calculation. Details differ: sometimes it is easier to work in conductances than resistances, etc. That's just flipping the fraction over.
The main difference between hand-calculation and SPICE is that a computer does math thousands of times faster than a human. So the computer may not set up a full analysis, it can just try-and-see. As I understand it, at some level of complexity, SPICE may assume the mid-point of a network is "half" the battery voltage. Then it computes all the currents and adds them up. By KCL they should sum to zero. They don't? Then SPICE adjusts its "half" guess, a little or a lot depending how far out it was, and does all the calculations again. In around a dozen to a hundred iterations it narrows-in on the "right" answer to 8 decimal places. (I may hand-calc 3 iterations to get 2 places, which is good enough for most work.)
Non-linear elements are represented by a formula which is recomputed for each iteration.
.TRAN analysis does all that many-many times over the specified time interval.
There's original FORTRAN SPICE, a C SPICE, and many folks who sell packages which include SPICE and typically with extra smarts (SPICE was put together quickly, so decades of smart programmers leads to little improvements).
Last edited:
In a previous life as a student, we had to program the underlying
algorithms ourselves ( in isolation, not everyting together) for educational
reasons b4 we were given access to spice.
We used this book :
< Computer Aided Network Design: Amazon.de: Donald Calahan: Fremdsprachige Bucher >
and this one
< Circuit Theory: A Computational Approach: Amazon.de: Stephen W. Director: Fremdsprachige Bucher >
The second one contained an awful lot of typos and misprints, looks like Wiley
could not afford a lector. :-(
In concept, it was excellent, but unusable without a buglist.
I bought that stuff in Nov. 76...
algorithms ourselves ( in isolation, not everyting together) for educational
reasons b4 we were given access to spice.
We used this book :
< Computer Aided Network Design: Amazon.de: Donald Calahan: Fremdsprachige Bucher >
and this one
< Circuit Theory: A Computational Approach: Amazon.de: Stephen W. Director: Fremdsprachige Bucher >
The second one contained an awful lot of typos and misprints, looks like Wiley
could not afford a lector. :-(
In concept, it was excellent, but unusable without a buglist.
I bought that stuff in Nov. 76...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.