Personally I do not understand why people do simulations of circuits that you can solder together in five to ten minutes
especially with inductors, and not one core is the same, and there are so many variables. Do you think it is some certification if the simulation happens to work? The reality test is reality. Simulations suck in many if not most cases, and that is why NASA is still simulating mars landings. Man years of work down the drain, or wiped harddisks.
Sure some filter programs will make it easy for you to find coefficients. And some circuits are cheaper to simulate than to build, but only if you have not enough clue to do the final design, or just want to play 'simulation'. Simulations are are insult to the power of the human brain.
Simulations should be made illegal, or at least taxed :-)
Here, here! I use the hell out of my old breadboard box/trainer/whatever they are called now :
formatting link
75839_-1
I surely did NOT spend that kind of money on it years and years ago, but a handful of parts and a bunch of the jumper wires and I am a happy man. It is a pretty good rule of thumb, that if it works on the breadboard, it will be even happier on a nice PCB. I have done more than my fair share of perfboard prototypes too. Nothing smells better than vaporized flux.
Used with care, simulations are good things. You can do Monte Carlo runs to check to see how well your circuit will stand up to component variations (multiply that 5 minutes by how much to do that in the real world?). You can probe points that are too high-impedance, or easily look at currents through conductors (or components) that wouldn't be convenient with a real world circuit. You can simplify parts of the circuit beyond belief (I often do this when investigating switched amplifiers where the 'real thing' will be driven from the PWM output of a processor).
You can also lead yourself to believe that a total piece of crap is a good circuit -- which is why simulations should only be a part of the design process, not the be-all end-all.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html
My brain can't do nonlinear quantitative analysis of complex circuits. Even if I wanted to, which I don't.
Simulation can be fun. You get to instantly play with values of resistors, caps, and inductors, which is hard to do on a breadboard. And you can do delay lines, ideal diodes, megavolt sources, all that. And you get pretty color graphs with huge precision.
I did this one last night:
ftp://jjlarkin.lmi.net/Peaker.jpg
Simulation is, if nothing else, a fast way to train your instincts. But I only sim simple subcircuits, never entire products.
I even use LT Spice to design and analyze DC voltage dividers. It sure beats banging on a calculator reducing Thevenin equivalents.
AC analysis can certainly make you think something is good when it actually has bugs. It is a good idea to ramp up a circuit just to see if it behaves well on start-up. For amps, feed a square wave, stuff like that.
If your temperature models are good, spice is far easier than chilling a circuit.
hat you can solder together in five to ten minutes
k?
ts.
This calculator may be of use. Solves voltage divider problems. Source voltage, current, resistance values, power dissipation, etc. I did it with Java Script many years ago.
Given a target ratio, it finds resistor pairs that we have in stock. It works by brute force, n-squared tests where n is around 1000 these days... runs plenty fast.
There must be a way. How about sticking in a voltage source, measure the resulting current, and then compute the resistor that would result in the same voltage in that position? That should work.
One I use most is to find a pair of E24 values to make a value:
grant@deltree:~$ ./rcalc5 782 Result Ra Rb Connect Error 780.0 16k 820R parallel -0.25% 780.0 390R 390R series -0.26% 780.0 510R 270R series -0.26% 780.0 560R 220R series -0.26% 780.0 620R 160R series -0.26% 780.0 680R 100R series -0.26% 780.0 750R 30R series -0.26% 781.6 2k7 1k1 parallel -0.05% 782.6 3k6 1k0 parallel +0.08% 782.8 5k6 910R parallel +0.10% 783.0 750R 33R series +0.13% 784.3 18k 820R parallel +0.29%
Never seems worth the effort to sit down and add more functions, unless I'm stuck for something, then I'll start LTSpice... Other times I'll add fluff like allowing 3k456 for value entry :)
grant@deltree:~$ ./rcalc5 3k456 Result Ra Rb Connect Error 3.449k 82k 3k6 parallel -0.21% 3.450k 2k7 750R series -0.17% 3.450k 3k3 150R series -0.17% 3.451k 30k 3k9 parallel -0.14% 3.452k 13k 4k7 parallel -0.12% 3.460k 3k3 160R series +0.12% 3.463k 91k 3k6 parallel +0.20% 3.467k 9k1 5k6 parallel +0.31% 3.470k 3k0 470R series +0.41% 3.471k 18k 4k3 parallel +0.43%
Grant.
Early days for me and spice, if I get really bored. Like today I want to power up a lightbox from a 32" LCD TV, 24V maybe 4A or 5A, would like to run it from battery, 21V to 30V boost/buck converter, design that then say nah. Go look for a 24VAC xformer and deicde on 3 x 7824s in parallel, worry about battery operation if the lightbox is good. Someone gave me me half a busted TV. No PS or main board. I figured a lightbox is the best I could get out of it. Thing's got 16 CCFL tunes in it, but they're too fragile to handle, make some fancy light fitting with.
Unless I think of a way.
Hell of a circuit, they use one 12V spec LCD inverter, multiplex that to eight parallel inverter transformers running from 24V, each driving two tubes. Unreliable according to web reports. Can't find the circuit, but it uses a cute P+N Mosfet pair in an 8pin SOIC, one for each xformer.
I couldn't see how to make a ramp, so yesterday I put a sinewave with offset, starting at 90' -- close enough for what I was looking for, I suppose it's in there help somewhere. Not read through it enough yet.
Spice is good in familiar territory, where one knows what to expect, but I'm less confident when going out side what I know with it. But one can waste much time with a soldering iron too, there's a balance.
On a sunny day (Fri, 22 Apr 2011 19:11:11 -0700) it happened John Larkin wrote in :
can solder together in five to ten minutes
My brain can do much more complex 'simulations' so can yours. It is just that the sellers of the simulation tools - basically a box of simple equations-, want you to believe their product is better than half a trillion neurons that perhaps even work at the quantum level, not even to mention things as cosmic consciousness, say interconnected brains. Even wondered why some inventions always happen at the same time, I mean somebody in US finds something, then somebody in Europe also had it in some paper shortly before or after, sort of a group awareness perhaps. Not only politics.
That is the right way to do it I am sure.
John, that is very beautiful, and I can do that trick in LTspice too. I do not know what you use that for though... Anyways sure, but like those toroids somebody was trying to simulate, I just got a whole bunch of electronics in yesterday, including many toroids, and some initial measurements show how many sorts of core material there are :-) Then there was somebody simulating 'LM317', I have measured some from different manufacturers and their transient response is all different.
So even if you simulation is 'perfect', the *only* proof is the real circuit with real components. And if it is just a 5 component circuit, and you HAVE those, then you may as well try it for real.
Von Braun had a mars plan (he had no simulations available to him), politics at that time found it too expensive, so now NASA is doing simulations and flying around the earth trying to see how long they can hold out up there... for a cost that is many times more than the original mars plan, while still not having brought back a single sample in all those years.
How many hours did you spend playing with simulation softwares (s intended), to make that simple filter? Are you SURE it will work OK with that inductor you have??? :-) Now you have to try it for real.
The left part of the circuit is just a test pulse generator, making a pretty-much Gaussian pulse. Note the 1-ohm LC filter, normalized for frequency but not for impedance. The thing I'm really simulating is between IN and OUT.
This is a simple pulse peak finder. The zero crossing of OUT flags the peak time of IN. I behaves like a differentiator but doesn't have the infinite noise multiplication that a true diff would have.
I spent less than an hour on this. I used FilterPro (V2!) to cobble up some very low-Q highpass sections, and experimented with how well they nailed the pulse peak. Spice lets me hary the Gaussian pulse width input and evaluate how much the OUT zero crossing walks.
This is like a constant-fraction discriminator, but without the delay line, and it's somewhat more tolerant of pulse width variation.
Although the simulations only give a theoretical look at things, they're great for playing "what if".
We used it for testing beyond-the-bench conditions to have a look at what happens to signals under those conditions.
Everything gets tested with real components eventually, but it helps to know where things are probably going to go in real life, especially when real life doesn't present those conditions when it's convenient to us.
The pulse function has a rise and fall time, so you can do a repetitive ramp that way. Or you can simple use a piecewise linear (PWL).
For analog circuits, it is a good idea to ramp the power supply up slower than you think will happen in real life and see if the circuit misbehaves. I'm just amazed at the amount of expensive electronics that thumps when shut down or turned on. These glitches are often problems in roll your own switchers. It is not the kind of thing somebody notices in real life, but for instance a big current surge when starting up can be a reliability issue. Undervoltage lockout circuitry can be way more complicated than most people suspect.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.