Do electrical engineers actually use Circuit Analysis skills?

[AchillesWrathfulLove]

Do electrical engineers actually use stuff that is taught in Circuit Analysis classes like Node Voltage Method, Mesh Current Method, Kirchhoff Voltage/Current Law, Source Transformation. All these usually only involve resistors. Are they somewhat altered in a way to analyze circuits with other electrical components?

As an electrical engineer what kind of laws/theorems/principles do you usually deal with in real life?

 

[anorlunda]

They do, but in the modern world more and more stuff is done by computers. So the skills are wrapped into the software.

You can say the same thing about add/subtract/multiply/divide skills because people use computers and calculators so much.

But an engineer who simply uses tools without understanding how and why they work will probably make huge mistakes. No employer would want him/her.

Stop complaining about the lessons you are asked to learn.

 

[Bonkers]

ooh, not sure you were complaining about having to learn stuff... ?
The KVL and KCL are the basis of Pspice, whether with just resistors and capacitors, or with semiconductors
However even the diode equations are quite complex, transistors are a lot worse, so we either use pretty basic approximations, or complex simulation or both.
A lot of design is really with sub-units, regulators, op-amps, where the individual parts are not considered (by the user) and "higher-level" parameters are used - like GBW, input offset, load regulation, PSRR etc.
I would look-up "cirmaker.exe" for a free very easy to use PSPICE simulation program.

 

[cabraham]

Yes.

Claude

 

[berkeman]

Yes.

+1.0

 

[berkeman]

Do electrical engineers actually use stuff that is taught in Circuit Analysis classes

Yes, every day. After a while you get good at looking at circuits and doing a ballpark analysis in your head, approximating operating points and gains and other things that interest you about the circuit. For more precise analysis, you use SPICE and Monte Carlo simulations (which vary component values some to look for accuracy problems when tolerances of components stack up).

You also learn more detailed circuit techniques to increase accuracy and to tolerate changes in temperature, etc. When you learn more about transistor circuits for example, you will learn important techniques to use in your circuit designs to maximize the performance. Tricks like emitter degeneration resistors and balanced current mirrors and cascode stages all are important tools that you will use.

Keep on learning, and maybe build a few electronics kits for fun and to start getting more practical experience with circuits. You will start to see things more intuitively as you look at schematics and electronic assemblies...

 

[jim hardy]

All these usually only involve resistors. Are they somewhat altered in a way to analyze circuits with other electrical components?

Yes. Resistors and DC analysis teach you the thought steps.
Next you will learn phasor notation and complex number arithmetic,
then solve circuits with resistors and capacitors and sinewave AC supply voltage/current.

Did i use circuit analysis ? Multiple times every single day.

In the real world when something doesn't work right and you're tasked with getting it running again you have to figure out how it's supposed to work.
That requires you to understand the circuit and be able to work it in your head.
Anything less is incompetence.

Heck , these days you need circuit skills just to keep your car lights working.

As an electrical engineer what kind of laws/theorems/principles do you usually deal with in real life?

Ohm Kirchoff Lenz Ampere Biot-Savot Gauss Carnot Bernoulli / conservation of energy
Being rather math challenged i always started with those basics and got a reputation as a good troubleshooter.

old jim

 

[jim hardy]

Good Luck in your studies, Mr Achilles. Work lots and lots of problems.

old jim

 

[AchillesWrathfulLove]

Thanks jim.

 

[Qurks]

Depends on what you do. If you do circuit design, you will be using KCL\KVL a lot. If you're using linearized components you can still do standard linear techniques but if not you're going to being plugging stuff into a numeric solver after doing first/second order approximations.

I personally haven't touched circuit stuff in 2+ years and it's very unlikely I will ever do so again but my area is not circuit design.

 

[Joshy]

I use it a lot.

 

[mpresic3]

I am currently learning microelectronics from a 1988 textbooks by Sedra/Smith and circuit theory by Hayt / Kemmerly, not for any course, but as a hobby. I was especially interested in a answer to this post given by professionals. It is reassuring that I am trying to learn something that is applicable and not just an idle hobby.

In this day and age where there are millions of transistors on a piece of metal smaller than a pie-plate, I was tempted to believe that nowadays, engineers no longer are interested in circuits with fewer that 100 components.

 

[Joshy]

We used Sedra and Smith at the university I attended not long ago and for many modules; they still use it, and I've met a lot of other engineers who use it as their reference. It covers a lot.

It depends on your field. I think your statement about < 100 components might be true for a lot of integrated circuits design with the exception of RFIC. Sedra and Smith seems to have lengthy chapters on transistors and using them for gain in IC design; I was surprised they covered power amplifiers and even non-linear class C. I think a lot of places outside of that don't think of transistors as a means for gain, but more as a switch focusing on R_ON, SOA, and the charge required to turn it on.

I found later chapters on amplifiers to be very helpful too although I don't see active filters used too often... maybe I'm too new.

 

[Bonkers]

A large proportion of the "shipped silicon" is in designs < 100 transistors - like all opamps and regulators, linear and switch-mode.
Larger designs will all use standard "cells" - like SRAM, DRAM, Flash, CMOS image sensors, - even full-blown VHDL uses standard logic cells - AND OR NOT etc.
So, there is always a need to massively optimise these little designs because they are repeated millions of times.

Active filters aren't so much used because of RC tolerances - the values you need for high order filters cannot be produced in volume.
switched-capacitor topologies solve this, but introduce switching noise. Most clever filtering is done in DSP where you can have any order you like.
One place real active filters DO get used is in low--IF radio architectures, where you can get the selectivity you need using opamps, rather than crystal filters. Polyphase is a good search term for this.

 

[Grinkle]

Do electrical engineers actually use stuff that is taught in Circuit Analysis classes like Node Voltage Method, Mesh Current Method, Kirchhoff Voltage/Current Law, Source Transformation.

Absolutely. It needs to be second nature for a circuit designer, like C syntax needs to be second nature for a C programmer.

 

[Grinkle]

They do, but in the modern world more and more stuff is done by computers. So the skills are wrapped into the software.

You can say the same thing about add/subtract/multiply/divide skills because people use computers and calculators so much.

I would phrase this differently. Computer software enables scale. When one says more stuff is done by computers, it implies less is done by the engineer. That is not the case in my experience. The designer does just as much analysis as they ever did, and at the same time they do more analysis using software.

I don't like the calculator analogy, because if one forgets how to do long hand square root solving, its not that big a deal. If one forgets how to apply Ohm's law, one is completely ineffective.

 

[analogdesign]

I would phrase this differently. Computer software enables scale. When one says more stuff is done by computers, it implies less is done by the engineer. That is not the case in my experience. The designer does just as much analysis as they ever did, and at the same time they do more analysis using software.

Interesting. My experience is completely different. I used to do a lot of hand analysis, but with modern nanoscale CMOS processes, the devices diverge so much from idealized square law equations that hand analysis is pretty much useless beyond very high-level system calculations.

Where we used to calculate bias currents and gm and stuff like that, now we just "characterize" the process by running a suite of basic simulations across device sizes to generate a bunch of plots. Then we size devices and current by picking point on a curve, graphically.

I would say I depend on the simulator far more than I did 20 years ago. I couldn't design a circuit in 28 nm, for instance, that worked without it.

 

[anorlunda]

EE is a very broad field. Only a very broad answer can be correct. That's probably true of almost all technical disciplines. They are much broader than students imagine.

When I was an EE senior, the popular question among my peers was "What do engineers actually do all day?" There was no good answer.

Many of the engineers here on PF do hands-on work with real life stuff. I was always on the analytical side making the simulation tools that others use. In most cases, I didn't have the slightest idea what those things I simulated look like except by pictures in a book.

 

[Grinkle]

hand analysis is pretty much useless beyond very high-level system calculations

I probably don't disagree with you. A couple things that occur to me -

1. This is a sliding scale and where one lands on it depends (among other things) on the experience of the engineer with a particular technology and circuit topology. I agree that for some level of precision, sometimes a pretty modest level, sims are required and hand calcs are not giving useable results. I also agree that all modern circuit design sooner or later crosses a threshold of requiring more than hand analysis to be confident one knows how the implemented circuit will actually behave.

2. High-level system calculations are a fundamental part of understanding ones circuit. If one can't do them, one can't interpret one's sim results. This is still analysis, and all the designers I know do a lot of it. I don't know any designers who can't relate their sim results back to device physics and circuit topology - I suspect that's because such designers don't end up being successful.

 

[analogdesign]

I probably don't disagree with you. A couple things that occur to me -

1. This is a sliding scale and where one lands on it depends (among other things) on the experience of the engineer with a particular technology and circuit topology. I agree that for some level of precision, sometimes a pretty modest level, sims are required and hand calcs are not giving useable results. I also agree that all modern circuit design sooner or later crosses a threshold of requiring more than hand analysis to be confident one knows how the implemented circuit will actually behave.

2. High-level system calculations are a fundamental part of understanding ones circuit. If one can't do them, one can't interpret one's sim results. This is still analysis, and all the designers I know do a lot of it. I don't know any designers who can't relate their sim results back to device physics and circuit topology - I suspect that's because such designers don't end up being successful.

I agree with both your points. I guess I was reminiscing on how different the design process is today. When I was in school (mid 1990s), we would work out the design of an op-amp, for instance, entirely on paper. We'd calculate the needed gm's and ro's, size the devices and choose all the currents using closed-form equations. Then we would have to enter in the circuit by hand in a text file and run it through SPICE. The simulations took a long time and the models weren't even that great, but we were able to make things that work.

Now, the simulations are super fast, and the models are empirical so you can't use them to get intuition anyway. We've even migrated to current density (aka the gm-over-i methodology) to size devices because the equations are a joke. So it makes sense to just iterate at the computer... put in an educated guess and tweak away! There are even optimizers now who will search the design surface for you but they make me nervous.

And you're right, folks who can't correlate simulations to physics end up doing something else.

 

[Qurks]

It's going to be real funny when machine learning and optimization algorithms take over the design process.

 

[analogdesign]

It's going to be real funny when machine learning and optimization algorithms take over the design process.

You're not wrong. Analog design automation has been a tough nut to crack but deep nanometer processes make it easier by greatly constraining the things you can do (for example, you can only choose discrete sizes, can only orient devices in one direction, etc).

I suspect it won't be too long until we have an analog synthesis and place and route flow, and then we will move up the level of abstraction in a similar way to digital circuit designers now mostly work with Verilog or VHDL. I consider myself in a sunset industry. Software will eat everything.

 

[Qurks]

You're not wrong. Analog design automation has been a tough nut to crack but deep nanometer processes make it easier by greatly constraining the things you can do (for example, you can only choose discrete sizes, can only orient devices in one direction, etc).

I suspect it won't be too long until we have an analog synthesis and place and route flow, and then we will move up the level of abstraction in a similar way to digital circuit designers now mostly work with Verilog or VHDL. I consider myself in a sunset industry. Software will eat everything.

I know, it's one of the reasons I avoided doing analog IC design. I also think there will only be a few SoC makers. Don't need multiple companies designing chips which do essentially the same thing.

 

[clope023]

Do electrical engineers actually use stuff that is taught in Circuit Analysis classes like Node Voltage Method, Mesh Current Method, Kirchhoff Voltage/Current Law, Source Transformation. All these usually only involve resistors. Are they somewhat altered in a way to analyze circuits with other electrical components?

As an electrical engineer what kind of laws/theorems/principles do you usually deal with in real life?

Yes, also lots of things can be modeled as resistor networks which follow KVL/KCL, which simplify the analysis.