EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: sarepairman2 on October 03, 2015, 08:06:26 pm
-
So is there a record circuit diagram showing the equivalent circuit for a passive component that just has a disgusting amount of non idealities?
Like iron core inductor from dc to daylight
-
Just about anything electrochemical in nature, (usually batteries, but also some chemical sensors) is likely to be very difficult to model accurately.
To a great extent it depends on your definition of a passive. If you strictly restrict it to fixed R, L, C you remove most of the more idiosyncratic devices.
-
Well i'd say that also equivalent models for passives in analog IC design simulator are pretty nasty in you get to deep sub-micron processes, for example a simple nwell/pwell resistor has a model with 3 terminals (r1, r2 and bulk) with a couple of capacitors and diodes (at least for .35 um AMS cmos process) then I would expect also that things like fingered, multi layer MOM (metal oxide metal) capacitors to have pretty complex models, unfortunatelly i doubt you'll find many schematics on internet, as most of this stuff tends to be covered by NDA's and even then mostly used and known just to the foundry and the cad vendor (most likelly cadence) to tune the PDK
-
The circuit equivalent for some of the ceramic capacitors should be interesting to work out. Not only do they vary their value depending on the applied voltage but they also have a piezo electric effect. Good luck modelling THAT in Spice :popcorn:
-
Anything you care to approximate can do, surely? Just crank up the precision..
It's not really a useful or meaningful question, I suppose? I mean, it's like asking, what's the longest English word? The longest non-technical word is, what, 26 letters, and the longest that was, I suppose, written down, is 1326 or something like that (the chemical name for, I think, the RNA or protein in some random virus). Such words are rather uninteresting and trivial in their composition: they are easily generated automatically from a chemical structure, following accepted rules (IUPAC or otherwise). In the same way, circuits are easily generated from certain rules. For example, one port of a ladder network is an implementation of a continued fraction, so that one can take a continued fraction of arbitrary length and compose a circuit out of it (or, for irrational numbers like sqrt(2) or pi, an infinite chain).
More generally, one can implement any finite differential equation in SPICE, though the space of models that will converge is quite a bit smaller than "all possible"...
Tim
-
> Just crank up the precision..
Exactly. Even a "plain" shielded cable with connectors in both ends can fill a whole page: I was once in a project where we took precision measurement wires through the same opening in an icebreaker's hull that the main engine power line were routed. We wanted to measure the events that caused the safety breakers go off. When a few ten kiloamperes are abruptly cut off, the generated electrical interference is, shall we say, non-trivial; especially when 'ground' is a few nautical miles away. First try, just routing cables: Fried electronics. Second try, modelling what is happening by two MS scientists, and adapting to that: Nothing broke, but no usable signal. Nth try, modelling by three professors and a bunch of other experienced people, determined too be the best result achievable within the bugdet: About 8 bits of usable signal out of 16.
-
Yeah, even for multiply shielded coaxial cable and various isolation and common mode methods, solving a problem like that can be non-trivial. That's probably about as much induced surge or EMP as any consumer product might be expected to handle due to nearby lightning strikes!
Tim
-
so this question promoted interesting stories