11
votes
\$\begingroup\$

I was wondering if anyone knows what the most widely (not favorite) used circuit simulation tools are in industry. I have a couple options here at the company I work at, but I still like to use LTSpice for most of my work. I'm pretty fresh out of college so I haven't had a very good chance yet to see what the industry is like.

\$\endgroup\$
  • 2
    \$\begingroup\$ I assume you are talking analog simulation. However there are tools for digital, RF and signal integrity simulation. Those are pretty much non-overlapping sets of tools. \$\endgroup\$ – Brian Carlton Aug 1 '11 at 21:40
12
votes
\$\begingroup\$

I'm going to sorta disagree with Olin.

If you're using a simulator for something you can do with a calculator and a piece of paper in a few minutes then you're using the simulator for the wrong purpose. If you assume that your 'napkin math' analysis will hold up in reality you're likely working on extremely simple, basic circuits in the first place.

More importantly you're really saying that it's OK to skip what is really the most important stage of pre-prototype design verification. This is a really bad idea if you're working on anything even moderately complex and very much can come back to bite you even on simple circuits. I've seen even the simplest IR transmitter oscillate due to parasitics.

Additionally, a huge use case of simulation that is a real pain to do with just a calculator is Montecarlo analysis. Almost every simulator supports this and it is very important for production designs.

It's actually very rare that a simulator will not give you more insight into a real circuit than a 2-minute, mostly intuition-based, analysis of the circuit will. A couple of hours generating the simulation can easily save you days waiting to get a prototype back only to find out that through some awkward component or parasitic interactions your theoretically perfect transmitter is just a lousy oscillator.

As an example, taken from an Analog Devices app note:

enter image description here

On the left we see a basic op amp circuit. On the right we see what this circuit would look like if one considers basic PCB parasitic effects.

No question that with 60 seconds and a calculator you can figure out what the circuit on the left is doing.

However, that is no substitute for producing a more complex model of the real circuit in the real application such as the circuit to the right. The right hand circuit is far from easy to analyze manually without hand-waving away components as irrelevant.

Additionally a proper simulation is going to use more realistic models for components, rather than the ideal models, which for a circuit of any complexity or speed is critically important to understand and analyze.

As to the original question:

Most circuit simulators are at least related to SPICE and many share a compatible or close to compatible model format. Additionally there are many other simulators which specialize in particular fields. Notably RF/microwave simulation, digital logic simulation, etc.

The most common simulators I've run into:

  • PSPICE - part of Cadence's OrCAD design package
  • Spectre - Mixed signal and RF simulator from Cadence (maybe the most common)
  • NI MutiSim - National Instruments simulation package
  • HSPICE - commerical SPICE implementation by Synopsis, also very popular
  • XSPICE - extended version of SPICE3, Altium uses this
  • SIMetrix - SPICE derived analog simulation

Which one(s) you will see in a given company is usually a function of their field of specialty (analog, mixed-signal, RF, etc), what integrates well in their chosen development environment and what they are historically comfortable with.

\$\endgroup\$
  • 6
    \$\begingroup\$ Your example is exactly the kind of abuse and over-reliance on simulators I was talking about. You can never quantify everything. Your example B is just another model for something that is infinitely more complex in reality. Good design is about knowing what really matters, and making sure that effects you didn't address don't matter. Unless you're doing something unusual, the whole idea is to chose reasonable values and use good layout so that example A is a valid model. At some point you have to actually know something, not just blindly enter data into a simulator. \$\endgroup\$ – Olin Lathrop Aug 1 '11 at 20:07
  • 5
    \$\begingroup\$ The whole idea is to create a model as close to reality as possible. Good design is about getting it correct and getting it reliably correct, not demonstrating what you can do without tools. This isn't a no calculator math test, its business. Any tool that increases correctness of your design and reduces costs should be fully exploited. Knowing what you don't know and how to test it is many times more important than thinking you know something and therefore not verifying it properly. \$\endgroup\$ – Mark Aug 1 '11 at 20:27
  • 8
    \$\begingroup\$ "The whole idea is to create a model as close to reality as possible." Wrong, and misses the point. The whole idea is to create a model that is useful. That means a model only complicated enough to describe the effects that actually matter. For most circuits, such a model can be solved with a calculator in a few minutes. For the nasty cases a simulator can be a useful tool, but with its own drawbacks. Excessive use of simulation means bad model and therefore poor understanding of what's actually relevant. \$\endgroup\$ – Olin Lathrop Aug 1 '11 at 21:43
  • 8
    \$\begingroup\$ Circuit Simulation is not a good design tool. However, it is an excellent verification tool. Think of it like having someone else check your work. It almost certainly won't catch every error, but it may catch some you missed. \$\endgroup\$ – Connor Wolf Aug 2 '11 at 2:31
  • 4
    \$\begingroup\$ .... I remember similar arguments about people blindly using calculators without undestanding arithmetic (and people do) and arguments about GPS SatNavs "making" people drive into rivers and off cliffs (and people do this too) but that's not the fault of the tool. Learn basic arithmetic, learn to read a map, learn to analyze circuits, then get a calculator, a GPS and a simulation tool. \$\endgroup\$ – MikeJ-UK Aug 2 '11 at 9:11
10
votes
\$\begingroup\$

I'm sure others will recommend particular circuit simulation packages.

However, circuit simulators are a tool, but usually not a very major one, and are certainly no substitute for using your own brain. Too often I see young engineers get so wrapped up in the third digit of some value coming out of simulator that they forget to think and actually understand what is going on instead of looking at fancy simulator output. The fancy output can only show you what is going on, not the much more important why.

When you really understand what a circuit is doing, then just about anything can be solved with a calculator in a minute or two. The understanding also let's you see corner cases that the simulator would only show you if you gave it just the right input. When you don't really understand a circuit, the fancy outputs will just give you a false sense of security.

Simulators have their place, but it's actually quite rare that they provide useful circuit analysis that a brain and a calculator can't come up with faster.

\$\endgroup\$
  • \$\begingroup\$ Excellent answer, +1. (Not that I would have expected anything less :-)) \$\endgroup\$ – stevenvh Aug 1 '11 at 17:35
  • \$\begingroup\$ This doesn't answer the question, -1. It should be a comment - a perfectly valid one \$\endgroup\$ – Toby Jaffey Aug 1 '11 at 20:19
  • 4
    \$\begingroup\$ @Joby: Sometimes a good answer addresses the real issue instead of what the OP literally asked about. \$\endgroup\$ – Olin Lathrop Aug 1 '11 at 21:44
  • \$\begingroup\$ +1, as this answer ought to be on top. The "counter-example", designated as B in the answer by @mark, is more of an "after the fact" model rather than the one that one comes up with at the outset. The key is to have enough design tolerances so that such stuff matters less, and that's where basic ckt theory and experience comes in! \$\endgroup\$ – Vaibhav Garg Aug 16 '12 at 8:46
2
votes
\$\begingroup\$

I have used SPICE in all it variants at a few workplaces.

Another format of simulator is SABER. this is used by Boeing, Airbus and various other safety critical industries. A subcontractor that I worked with was designing DC/DC converters and required DO-254 certifications (aviation) and their customer (and certifying body) required SABER models of the circuit. To this end we got a SABER expert in to peform the simulation work and used a SPICE variant for our inhouse design!!

I think there is a third major format for circuit simulation.

\$\endgroup\$
2
votes
\$\begingroup\$

I use LTSprice from Linear Tech. Free, easy to add libs from other vendors and real beauty compared to other free packages such as TINA from TI.

\$\endgroup\$
  • \$\begingroup\$ The question is not asking what you personally use. \$\endgroup\$ – endolith Aug 3 '11 at 19:07
  • 2
    \$\begingroup\$ @endolith well.. thanks for pointing that out in style. \$\endgroup\$ – Frank Aug 4 '11 at 2:26

Not the answer you're looking for? Browse other questions tagged or ask your own question.