What approach should be used when solving a circuit?

[Boltzman Oscillation]

TL;DR Summary

I want to know what methods are typically made in analyzing a method.

I am close to graduating as an EE major but I have never been able to organize a step by step method on analyzing a circuit. It seems to me that every time I am trying to analize a circuit I end up with a bunch of equations and nothing more. I know that I should:

1. Know what I am solving for (Vout, Vin, Pout) and find equations in relation to that.
2. Use KVL, KCL, and Ohms law.
3. Note the number of unknowns in my equation and try to find that same number of equations.

Now using these three rules then I should be able to find anything I want in a circuit but usually I dont. Let us say I need to find Vout for a circuit. I usually find an equation, any one, that has Vout in it. I then find other equations using KVL and KCL but only enough to match the number of unknowns. I then end up with a bunch of equations where I cannot single out Vout to the left side of the equations like so:

$$V_{out} == right side$$

What should I be doing to actually get a good final equation?

 

[phinds]

What should I be doing to actually get a good final equation?

It's going to depend on the circuit but one important step you left out is to simplify the circuit before you start doing loop equations or node equations.

Also, focus on a step by step approach, not trying for a final equation all at once.

Thevinin and Norton are your friends.

 

[jasonRF]

I agree with phinds - simpify the circuit first, and follow a step-by-step approach.

To be more explicit, instead of focusing so much on the one quantity you want ($V_{out}$), notice that you need to setup equations that can be solved for all of the currents/voltages in the circuit. There are a couple of standard approaches: nodal analysis and mesh analysis. I'm guessing whichever book you used for basic circuit theory covered this - have you re-read the relevant sections of your circuit book(s)?

The idea is that you follow a systematic procedure and end up with a system of linear equations $\mathbf{A x} = \mathbf{b}$. Then you basically just solve for the unknown voltages and/or current contained in the vector $\mathbf{x}$. There are situations where forming one or more of the equations takes extra work, but it is still a systematic procedure. There are a number of textbooks that describe these things better than I ever could, so again I recommend looking at your books. If you end up with specific questions please feel free to post and I'm sure folks here will be happy to help.

jason

 

[Boltzman Oscillation]

I agree with phinds - simpify the circuit first, and follow a step-by-step approach.

To be more explicit, instead of focusing so much on the one quantity you want ($V_{out}$), notice that you need to setup equations that can be solved for all of the currents/voltages in the circuit. There are a couple of standard approaches: nodal analysis and mesh analysis. I'm guessing whichever book you used for basic circuit theory covered this - have you re-read the relevant sections of your circuit book(s)?

The idea is that you follow a systematic procedure and end up with a system of linear equations $\mathbf{A x} = \mathbf{b}$. Then you basically just solve for the unknown voltages and/or current contained in the vector $\mathbf{x}$. There are situations where forming one or more of the equations takes extra work, but it is still a systematic procedure. There are a number of textbooks that describe these things better than I ever could, so again I recommend looking at your books. If you end up with specific questions please feel free to post and I'm sure folks here will be happy to help.

jason

So you are saying that I should be using either nodal or mesh analysis? I find nodal/mesh to be difficult to do when I am given a component such as a BJT. If I try nodal then I have to find a way to transform my voltage sources to current sources which I can if I have a handy resistor in series to it but that is almost never the case. Mesh analysis requires there to be only voltage sources (the way I've learned it) but once again, the BJT is modeled by a dependent current source that needs to be converted into a voltage source; I figured I could just use $r_{\pi}$ to convert it since it is in parallel with the current source.

 

[LvW]

So you are saying that I should be using either nodal or mesh analysis? I find nodal/mesh to be difficult to do when I am given a component such as a BJT.
....

Here is a "trick" which can help to simplify BJT amplifier analyses:
For DC operating point calculations you, normally, are assuming a fixed DC voltage between base and emitter of VBE=0.7 volts (or any similar value).
For calculation purposes, you can replace the base-emitter path with a DC voltage source of the same value - and make use of the superposition theorem. Thus, you are splitting the calculation into two separate parts which reduces the number of equations which have to be combined.
This procedure is allowed because of the existence of the substitution theorem.

 

[gleem]

In the text " The Art of Electronics" by Horowitz and Hill there is little formal network theory and need for solving large systems of equations. They believe that their approach gives a better understanding of the choices and compromises that engineers face in circuit design.

The authors develop models for active devices as BJTs and FETs. They avoid what they call unnecessary pedagogical approaches not used by practicing engineers. They look at circuit design as more of an art using a combination of basic laws, rules of thumbs, and a large bag of tricks. They do not get lost in math.

Check it out.

 

[jasonRF]

So you are saying that I should be using either nodal or mesh analysis? I find nodal/mesh to be difficult to do when I am given a component such as a BJT. If I try nodal then I have to find a way to transform my voltage sources to current sources which I can if I have a handy resistor in series to it but that is almost never the case. Mesh analysis requires there to be only voltage sources (the way I've learned it) but once again, the BJT is modeled by a dependent current source that needs to be converted into a voltage source; I figured I could just use $r_{\pi}$ to convert it since it is in parallel with the current source.

Well, you asked a very general question, and mesh and nodal analysis are two of several general methods. As you mentioned, you can use the fact that the controlled current source in the hybrid-pi model is parallel to $r_0$ (I believe $r_\pi$ is in the other leg) and transform it to a voltage source. Also, there might be a source transformation that applies to your pesky voltage sources that you might not be aware of. An example transformation is below (from Basic Circuit Theory by Desoer and Kuh):

without seeing a specific circuit we can only talk in generalities.

EDIT: also, from your post you mention needing a resistor in series to transform a voltage source to a current source. Just in case you don't know, Thevenin-Norton theorem applies for general impedances, not just resistances.

Jason

 

[Boltzman Oscillation]

In the text " The Art of Electronics" by Horowitz and Hill there is little formal network theory and need for solving large systems of equations. They believe that their approach gives a better understanding of the choices and compromises that engineers face in circuit design.

The authors develop models for active devices as BJTs and FETs. They avoid what they call unnecessary pedagogical approaches not used by practicing engineers. They look at circuit design as more of an art using a combination of basic laws, rules of thumbs, and a large bag of tricks. They do not get lost in math.

Check it out.

I have the book but stopped reading because I the solution manual does not exist. How can I make sure I am doing the circuits correctly? If I just solve the problems on my own without being corrected then I might just end up practicing a bad habit. Should I just read it without doing the problems?

 

[Jody]

Why not run the circuit through a simulation software such as LTSpice?

Something else that helps me is trying to solve the circuit with multiple methods if it's practical enough. The answers should be the same if it's right.

 

[gleem]

I have the book but stopped reading because I the solution manual does not exist. How can I make sure I am doing the circuits correctly? If I just solve the problems on my own without being corrected then I might just end up practicing a bad habit. Should I just read it without doing the problems?

Do you have the accompanying "Student Manual for the Art of Electronics"? It has more worked examples and solutions to select problems.

Many problems in the book are design problems for which he gives the specs. Does your design meet the specs. Other problems are of the show variety or prove variety. He gives a lot of information and builds on it. Design a circuit to do this, using the information and considerations that were discussed previously maybe all the way back to the beginning of the book. He gives lots of examples to build on.

Don't just read, study the book. Understand the reasons for the discussions. There is a boat load of information in that book.

 

[yungman]

Since you are close to graduation, I am going to assume you are going to enter the job market soon, so instead of talking about "how" to make the equations to calculate Vout, KVL all that, I am going to talk about the more practical way in real life on the job after almost 30 years mostly design analog, RF, IC circuit.

First off, I don't recall even ONCE I wrote KVL type of thing, I don't think I ever even sat down and wrote these kind of stuff at the beginning of a project. Now don't mistaken that I don't think theory is important, it's very important particularly now a days in microwave that you need all the knowledge of electromagnetics, signal integrity, RF designs etc. that require a lot of theory. BUT, analyzing using KVL and writing those equations in the books is NOT one comes to my mind.

When I approach a new project say to amplify a signal.
1) I look at the input drive characteristics to my circuit ( drive capability, impedance, voltage range, speed etc.). Then I look at what am I suppose to do with it. Then I look at the output requirements ( voltage range, slew rate, current capability etc.)

2) After those are clear, I would immediately think of what type of circuit that can do the job. I would google and look at other examples that do the similar job. I would go from the easiest way...Like using an opamp I can buy.

3) To see whether I can use an opamp, this is where a lot of time is spent going on Digikey, Mouser etc. type of distributors to look. It is a learning process how to look at requirements of choosing to find something that fits your needs. Then it's a long process to read the datasheets of different opamps. Look at input range, input current, frequency response, slew rate, output range, output drive etc. etc. To confirm it can do the job.

4) If you can find one that can do the job, consider yourself lucky. If not, you might have to go to transistor circuits. You need to come up with the rough design ( no KVL and all that). Then repeat step 3) above on finding the right transistors. If transistor doesn't work, you research another way until you find one.

5) After determine what to do, then you start drawing up the circuit, look at the DC bias ( still I never use KVL stuffs). Circuits are usually the easy and simple part, believe it or not, circuit in the books usually works. If you find yourself having to have fancy tricks in the circuit or very critical in resistor values and all, IT'S LIKELY A BAD DESIGN. almost all the end circuits are very simple just like in the textbooks. The simpler the better, too complicate, look inward that your design is NOT GOOD. If you need KVL to find answer, it's time to look at your design.

Circuit design is the easy part, in my experience, making the circuit stable and reliable, not susceptible to interference, not radiate noise to the surrounding is where you earn your keeps. this is the real life hard part. This is too long to talk and off the topic of this thread. There is a big difference between academic environment and real life on the job. I've seen people from prestige colleges with good grades didn't manage to transition from academic to real life.

It's too long to talk about this subject, I can only give you a specific example above how to approach a task from my experience.

 

[Jody]

I feel like I might be in similar shoes with OP if they are graduating soon. I'm not sure how jobs worked 30 years ago, but all 3 of my jobs (including an RF engineer position) within the last 2 years and a few months required the fundamentals including KCL and KVL. I would definitely say at the very least almost every interview was loaded with circuit analysis questions, and I have done a lot of interviews.

 

[yungman]

I feel like I might be in similar shoes with OP if they are graduating soon. I'm not sure how jobs worked 30 years ago, but all 3 of my jobs (including an RF engineer position) within the last 2 years and a few months required the fundamentals including KCL and KVL. I would definitely say at the very least almost every interview was loaded with circuit analysis questions, and I have done a lot of interviews.

I have my share of interviews not 30 years ago, more recently. They asked like what is the DC setup and all, but it is a lot more obvious that you don't have to get into KVL notes. It's like if they ask you a question of a BJT bias, you look at the voltage at the base, emitter is 0.7V below for NPN. You know emitter is low impedance, it's like a voltage source, you easily calculate the current if there's a resistor to the -V. Point is it's very simple, you don't need to set up equations.

KVL in school get into a lot of resistor divider network, looking at the voltage at a note, combination of current and voltage source. In real life circuit, you avoid design like this at all cost, you want it to be simple, one stage at a time. If you have to fix in all the stuffs at one time, that's bad design.

Particular in RF circuits, each stage is very small, one RFIC, one or two transistors. In is very simple, the difficult part is the impedance with distributed elements using stripline and all. Not the normal KVL stuffs.

My interviews on RF is more on Smith Charts matching impedance type of thing. Now, that's important. not KVL!

 

[Jody]

I would put a lot of money on it, that interviews for entry-level candidates are very different from people who have more than 10 years of experience. Almost all of my interviews did involve setting up equations and using fundamentals.

 

[yungman]

I would put a lot of money on it, that interviews for entry-level candidates are very different from people who have more than 10 years of experience. Almost all of my interviews did involve setting up equations and using fundamentals.

That I don't know, My entry level was back in the 70s and 80s, but still, the question is more giving you a practical circuit like an opamp with common mode voltage and tell you to give the voltage on each node. BUT, it should be very obvious with knowledge to use KVL. I had like transistor bias circuits and asked for DC. As I said, you know the circuit, pick what is the low impedance note and not resort to KVL type of calculation of loops and all that.

 

[yungman]

Let's put it this way. I was a manager of EE of like 12 years. If I ask the candidate about the voltage of a circuit like an opamp or transistor, if the candidate tries to write KVL, he/she will NOT hear from me again. If they don't have the common sense to answer the question, they DON'T have it.

If someone don't have the common sense and said to me " in the emitter if current the transistor is 10mA and the r'e is 2.6ohm, so I have to calculate this in the divider network. I don't need this kind of people anywhere close to my group. This is real life. And don't give me how in the physics why it's 2.6ohm. I'll walk the candidate out the door.

 

[DaveE]

I have the book but stopped reading because I the solution manual does not exist. How can I make sure I am doing the circuits correctly? If I just solve the problems on my own without being corrected then I might just end up practicing a bad habit. Should I just read it without doing the problems?

If you want to be successful in a EE design position, you need to learn ASAP how to check your work and know if you are correct without help. Your coworkers are not being paid to solve the same problem as you, they have their own work to do. In fact, in may jobs, there isn't anyone that knows how to check your work. Yet, the final product does have to work, managers are not very tolerant of repeated inefficiencies or failures.

 

[yungman]

If you want to be successful in a EE design position, you need to learn ASAP how to check your work and know if you are correct without help. Your coworkers are not being paid to solve the same problem as you, they have their own work to do. In fact, in may jobs, there isn't anyone that knows how to check your work. Yet, the final product does have to work, managers are not very tolerant of repeated inefficiencies or failures.

It's the result. I don't want to hear reasons why, I just want it to work...AND don't make me having to worry about it.

Bottom line, I don't give a crap why it doesn't work, make it work. You can use voodoo, just make it work.

 

[DaveE]

Many EE programs don't really address your question. Yes KVL, KCL, Thevenin, etc. are necessary tools, and you will use them sometimes. However in practice much of EE design work is actually more like familiarization with common problems and solutions; pattern matching and modeling with similar solved problems. We don't "reinvent the wheel" very often, we are paid to get a good solution as quickly as possible. Sometimes that requires creativity and new designs, but it is usually just understanding what is already been done by others and modifying it to suit your needs. The crucial step is that you have to be able to really understand other's designs so you know how to use or modify them.

I'm pretty out of touch with entry level positions these days. But, in the past, most of these skills are learned in entry level position where you are learning from more experience engineers "on the job".

 

[yungman]

Many EE programs don't really address your question. Yes KVL, KCL, Thevenin, etc. are necessary tools, and you will use them sometimes. However in practice much of EE design work is actually more like familiarization with common problems and solutions; pattern matching and modeling with similar solved problems. We don't "reinvent the wheel" very often, we are paid to get a good solution as quickly as possible. Sometimes that requires creativity and new designs, but it is usually just understanding what is already been done by others and modifying it to suit your needs. The crucial step is that you have to be able to really understand other's designs so you know how to use or modify them.

I'm pretty out of touch with entry level positions these days. But, in the past, most of these skills are learned in entry level position where you are learning from more experience engineers "on the job".

That's what the newly graduates don't get, that's the reason I am so harsh in this. I have seen people from UC Berkley and Standford that I would fire if they were in my group. The last thing I want to hear is WHY it doesn't work. MAKE IT WORK.

You talk to me about KVL, I'll walk you out the door.

 

[DaveE]

It's the result. I don't want to hear reasons why, I just want it to work...AND don't make me having to worry about it.

Bottom line, I don't give a crap why it doesn't work, make it work. You can use voodoo, just make it work.

Absolutely. But, I also need you to give me confidence that it will work tomorrow, too. Or that if there is a problem no one will die or go bankrupt as a result. "Working" has a complex definition.

Building trust is also necessary. Mistake can be more significant to your career than they initially may appear. OTOH, it would (probably) build trust if you came to me and said "I found a problem with my design, these are the ramifications, and this is how I will fix it." No one's perfect after all.

 

[yungman]

Absolutely. But, I also need you to give me confidence that it will work tomorrow, too. Or that if there is a problem no one will die or go bankrupt as a result. "Working" has a complex definition.

Building trust is also necessary. Mistake can be more significant to your career than they initially may appear. OTOH, it would (probably) build trust if you came to me and said "I found a problem with my design, these are the ramifications, and this is how I will fix it." No one's perfect after all.

That's later on, I am talking about the interview. It won't pass my interview...

I would test them with more common sense questions. I have my sets of question...not including KVL. I have to write up my questions that I actually gave to the jr engineers I interview. I actually don't talk to them until I see the test results. If they fail, I just politely talk a little and walk them out. When they cannot answer, I would talk to them and ask " what would you do if this ...or that". I want to see how they face a problem. It's the thinking process I am looking for.

My question is only in AA level, you'd be surprised how many so called "degreed engineers" failed to answer the questions for like a two years Heald college stuffs.

 

[Jody]

This seems to infer that the people who couldn't answer your questions could answer it using KVL or KCL. My guess would be that if they couldn't answer your questions, then they probably couldn't do it with KVL or KCL neither.

My interviews were loaded with circuit problems. Similarly: They might start me off in an empty room with a 30 minute exam on paper before talking to me; they'd have my work in their hands during they actual interviewing portion where you talk to a person. They would definitely ask for the current and voltages at intermediate nodes as well, and I saw it come into play in my positions.

I normally applied for design positions; also normally in R&D groups. My past jobs I was in one large company that I don't think a lot of people have heard of although they've seen their products in the news very often. That job was basically all Kirchhoff's circuit laws. The other two are the Behamoths of their industry. KCL and KVL isn't an everyday thing for those two roles, but specifications are so tight, that we're looking for every knob we can turn in order to succeed.

 

[DaveE]

My interview questions for analog design were always very simple schematics with "canonical" behavior. Like step response of an LC tank with Q=∞. I'll draw a current mirror, you explain it. Frequency response of a dominant pole compensated op-amp (integrator, or amp, whatever). Draw a flyback DC-DC converter and explain how it works. etc.

If they really understand the basics, then I want to know about the more complex stuff that they have knowledge of, a school project or such. If you don't know the simple stuff, I won't teach it. But if you do know that stuff, and you've shown me that you did sometime else worthwhile, then I'll probably assume that you can teach yourself the details you need. After all, no one would come in already knowing exactly what they'll be doing in the future.

BTW, it was astounding to me how many EEs can't deal with the step response of an LC circuit. There isn't a physics student out there that can't describe the motion of a weight hanging on a spring when you pull it down and let it go, yet most EEs can't do the same with the circuit version. Then sometimes you get a ringer, that just writes down the whole answer, frequency, amplitude, characteristic impedance and all, because they've used it before.

 

[yungman]

I gave very simple test. But you'd be surprised how many people failed.

Just very simple question like these. fill out the DC voltages and currents, then draw the output waveform respect to the given input waveform.

If anyone needs KVL, there's the door. Of cause, they can use calculator.

 

[yungman]

I want to get off the interview and want to advice EE students. Study hard on Electromagnetics. It is HARD, but it is getting more and more important in the new age of RF and mixed signals. Learn how to do pcb layout ( IC layout also if applies). The topic of Signal Integrity is VERY VERY important. I don't mean learning any layout software, I mean the concept of layout, what is important, how to control the ground current return paths, component placements and trace layout. I layout all my own boards as I worked in mixed high speed and digital circuits. You cannot rely on pcb designers to do that for you. PCB layout is where the battle win or lose.

Even if you are not going into RF design, it is very important to know signal Integrity which needs the knowledge of electromagnetics. EMI emission and susceptibility is very important when RF coexist with low noise and digital circuits together. Learn how to layout for signal integrity. Study on closed loop stability, Bode Plot. These are where the rubber hit the road. Circuit design is very simple in general, it's how to make it not sing and don't make other circuits sing is where you earn your keeps in the job. People learn to draw ground return in schematic with just a triangular symbol, that little triangle is more than half the battle in EE. Circuit design is very simple, you take care of the grounding, stability, signal integrity, circuit is really as simple as you see in textbooks most of the time...except RF, then it's a different world. The world of dancing on the Smith Chart. That's where layout is even more important when you start using stripline and microstrip as distribute components.

 

[LvW]

I would test them with more common sense questions. I have my sets of question...not including KVL.

Concerning "common sense question": When you ask "tell me how the collector current of a BJT is controlled", I think that AT LEAST 50% of the beginners will answer "base current" because the only equation they remember is Ic=B*Ib (and because many books are wrong in this respect).
But they will fail completely when you ask them to explain this phenomenon.

 

[LvW]

By the way - the late great Barrie Gilbert wrote:
"...every time I undertake a new ...project I start out with just four basic truths

* Like elements match well
* V=I*R
* dV/dT=I/C
* Ic=Is*exp(Vbe/Vt) for bipolar out of saturation (Ids=K(Vgs-Vth)² for MOS in saturation).

 

[yungman]

By the way - the late great Barrie Gilbert wrote:
"...every time I undertake a new ...project I start out with just four basic truths

* Like elements match well
* V=I*R
* dV/dT=I/C
* Ic=Is*exp(Vbe/Vt) for bipolar out of saturation (Ids=K(Vgs-Vth)² for MOS in saturation).

I kind of regret I did not stay longer in the IC designer job. In my days, we mainly did custom bipolar IC, plenty familiar with Ic=Is*exp(Vbe/Vt), I never even worked with (Ids=K(Vgs-Vth)².

Yes, "Like elements match well" is VERY important in IC design. absolute resistor value inside IC was like 30% those days, but if you put two resistors side by side, they match very well, it's all about layout in the IC, the direction, the transistor geometry/orientation... That's the reason I advice people to learn layout, be it pcb or IC. Layout is EVERYTHING. It all start with component placements. Good component placement make the ground and power plane easy and complete. Also making trace connection simple.

Lately I have been designing very high end hifi power amps as hobby, I looked at the service manuals of some pcb of the amps, I saw traces running all over the place, long and thin. Their layout really sucks. A good layout makes a complicate circuit look like very simple, very few traces and traces are short. Long trace forms large current loops...like a loop antenna for UHF, ready and willing to receive EM noise. Also emitting noise to the surrounding.

 

[yungman]

Concerning "common sense question": When you ask "tell me how the collector current of a BJT is controlled", I think that AT LEAST 50% of the beginners will answer "base current" because the only equation they remember is Ic=B*Ib (and because many books are wrong in this respect).
But they will fail completely when you ask them to explain this phenomenon.

I would make them design so Ib is not even important and not even worry about it. Like when designing a circuit, make sure resistance of the bias network for the base is not too high that it will affect the voltage at the base. Then I don't even worry about Ib, hfe all that, that I can assume base current is 0, hfe is high.

If base current is an issue cannot be ignore, it's time to think MOSFET or jFET. That's the reason I am against thinking KVL. You use common sense to eliminate complication. If base current effect cannot be eliminate, change transistor! If someone stuck with trying to KVL to design, that's bad. Yes, there are cases Ib is a big issue, like transimpedance amps that measure current into pA. there's no way to avoid Ib...more the drift of Ib. You go to MOSFET, no KVL is going to make the day with BJT, go MOSFET...with new sets of problem!

 

[Jody]

Quite honestly I think you'd be pointing me to the door and I'd probably be fine with that, but I'm curious why you're so against KVL or why it wouldn't work well? I think it would. I personally would not be discouraging recent grads from using it in interviews; I'd actually encourage it and would prefer if my coworkers could solve problems using the fundamentals.

 

[yungman]

Quite honestly I think you'd be pointing me to the door and I'd probably be fine with that, but I'm curious why you're so against KVL or why it wouldn't work well? I think it would. I personally would not be discouraging recent grads from using it in interviews; I'd actually encourage it and would prefer if my coworkers could solve problems using the fundamentals.

You look at my questions and comments? How you can using common sense to simplify the problem?

The idea is using common sense, simplify, knowing what is important, how to eliminate unnecessary complication. It's the real world in the job. I don't care how you do it, just get there, fast, reliable.

I've seen people from UC Berkley and Standford, they just so hung up with useless things, I would fired them if they were under me. RESULTS.

That's the problem, so many people studied and get good in all the formulas, but they don't have what it takes to design, getting the job done. That's the one I want to eliminate. Believe me, there are plenty of those people. Some people just don't have it, I just want to find people that have it.

You said you are into RF, I would test you in a totally different things. Smith Charts impedance matching, how to design distribution elements.

I think today, it's the job seeker's market, they let you pass with simple questions. I was applying for jobs in 2002, the deepest of depression in hightech. You better believe I got test on Smith Charts impedance matching on the spot. What is KVL?

 

[yungman]

Quite honestly I think you'd be pointing me to the door and I'd probably be fine with that, but I'm curious why you're so against KVL or why it wouldn't work well? I think it would. I personally would not be discouraging recent grads from using it in interviews; I'd actually encourage it and would prefer if my coworkers could solve problems using the fundamentals.

Ha ha, I hope if you work on my two questions in post 25, you don't ask me about what's the beta of the transistors and doing KVL. I would hope you would ask me to lower the 20K and 10K voltage divider resistors to get better consistence.

These are REAL circuits in projects, not some made up stuffs.

 

[Averagesupernova]

Ha ha, I hope if you work on my two questions in post 25, you don't ask me about what's the beta of the transistors and doing KVL. I would hope you would ask me to lower the 20K and 10K voltage divider resistors to get better consistence.

These are REAL circuits in projects, not some made up stuffs.

Well I might be one you'd show the door too. And knowing what I know, I wouldn't be sorry about it. I'd likely ask the beta to determine if it's necessary to improve the voltage divider bias as you describe. Your management style would attract fly by the seat of their pants type folks in my opinion. Lowering the voltage divider resistance also draws more power. Would you show me the door if I asked if this circuit went into a product that is powered by battery? Concerned about battery life and all. You want an engineer who is actually able to make the best set of compromises or one that simply gives you answers that make you feel good?

 

[hutchphd]

If you only carry a hammer in your toolbox. every problem will look like a nail. The expert carpenter carries a panoply of tools, and works to become better with each of them..

 

[yungman]

Well I might be one you'd show the door too. And knowing what I know, I wouldn't be sorry about it. I'd likely ask the beta to determine if it's necessary to improve the voltage divider bias as you describe. Your management style would attract fly by the seat of their pants type folks in my opinion. Lowering the voltage divider resistance also draws more power. Would you show me the door if I asked if this circuit went into a product that is powered by battery? Concerned about battery life and all. You want an engineer who is actually able to make the best set of compromises or one that simply gives you answers that make you feel good?

There's FETs.

That was a test question. The idea is you use common sense approach, not just analyze. I had very good luck in hiring people.

 

[Averagesupernova]

Of course there are FETs. Or any other of hundreds of ways to accomplish something. I find your approach is silly. Ask an open ended question, expect an open ended answer.

 

[berkeman]

It has been suggested that the EE interview posts be broken out of this thread into a separate thread. What do folks think?

 

[jedishrfu]

I would put a lot of money on it, that interviews for entry-level candidates are very different from people who have more than 10 years of experience. Almost all of my interviews did involve setting up equations and using fundamentals.

No kidding unless you get a stupid interviewer. I once nterviewed for a senior level position at a startup and got asked how to swap two variables. To be fair he was a junior programmer who didnt know how to interview.

Also interview questions don't tepresent the job either but are given to see how you react under pressure, how confident you are and sometimes how creative out of the box you are. It always good to collect names of interviewers for thank yous and more importantly interview them about the job working environment and anything else you can think of. Try to build a connection.

On another interview, i got asked a trick C++ question and before answering I asked the interviewer if he read Dr Dobbs. I answered it and then countered with one frim the Dr Dobbs article on the top ten C++ interview questions and he couldn't answer it so I gave him the answer.

I turned the job down because they wanted me to fill a project mgmt position, promising to transition me to development. Bottomline, don’t fall for the promised switcheroo and negotiate your salary as leaving one job to go to the next often means a loss of vacation time and other intangibles.

 

[Jody]

I think a thread related to interviews may be of high interest (it is to me) and there are a few interesting perspectives buried in this thread, which really should have been more about approaches to solving problems or analysis. I think it got mixed up because of course we're asked technical questions during the interview.

 

[yungman]

I used to give the test like in post #25 to technicians. These are something I learned in Heald college for AA degree. The book is by Malvino. I thought people should be able to answer easily even if they only gone through AA degree...WRONG. Then I put these type of question for engineer and thought this is only to eliminate the fake ones....WRONG! You'd be surprised how few can answer those.

I NEVER had a formal EE education. My major was Biochem for under grad. Electronics was and IS my passion and hobby. I completely self studied. I went through Heald in 1979 and got my AA in electronics in 9 months. I never even had the class on analog, I just read the Malvino book. Then change to different jobs to learn different facet of electronics like with LeCroy to design data acquision, Exar to design IC, Seimens to design Utrasound medical scanners etc. I expect people that went through the analog class would find it a breeze to answer those questions! I studied back ALL the undergrad EE and beyond....Like Electromagnetics, calculus like Differential Eq and Partial Diff Eq in order to study EM and RF. I thought people that went through college will have no problem with these...AGAIN...NOT...What are they teaching in college? All I was looking for was someone with common sense problem solving ability.

I had a job interview in 2002 with the same type of question...as Sr. EE. I got hired, we talked later, we were laughing how these kind of simple question can trip up those experienced EE.

 

[Averagesupernova]

All I was looking for was someone with common sense problem solving ability.

That ability in my opinion is more part of a person's core personality. It has little to do with what is taught in schools. School may unlock what a student didn't know they had when it comes to that ability but I think that's an exception.

 

[yungman]

That ability in my opinion is more part of a person's core personality. It has little to do with what is taught in schools. School may unlock what a student didn't know they had when it comes to that ability but I think that's an exception.

You've read my questions in post #25? It's very simple. How can it trip up so many people? These are real life circuit one encounters in analog designs. These questions are straight out of this book which I had to 1979 version. This was used in Heald College ( a technical 2 years AA degree for technicians). :
https://www.amazon.com/s?k=malvino+electronic+principles&crid=OHW65HC5U21O&sprefix=malvino,aps,351&ref=nb_sb_ss_i_3_7&tag=pfamazon01-20

I don't know how we end up talking about interviews. The original question is about "how to analyze...KVL...". In my view and from my experience specialized in analog, RF, mixed signal designs, I never ever use all the network theorems like Thevenin's Theorem, KVL and all those. Just look at the circuit, make it simple like in my two questions so the voltages in every point is very obvious with V=IR.

If I really want to go deep, I rather looking at the real circuit, how the ground return current run around on the ground plane, how to avoid ground loops that generate EMI. You really want to talk deep circuit theory, this is deep. And this is really really important in real world. That's what I was contracted to KLA Tenco in 2002 to help them on the layout of their few Giga pixels CCD circuits( when the commercial camera was only about 6M pixels). In school, the ground return is just a triangle and magically return the current. In real life, this is the major issue in circuit design. You get that right, the circuit is as simple as in the drawings in the textbook.

And in RF circuits, Smith Chart impedance matching. Yes, I was actually tested when I applied to Maxim in 2003. Too bad it took them like 3 months before they decided to make me an offer, that I long found another job already. The guy told me to hold on, the offer would come soon and he was angry when I said I got a job already! That was to design the reference circuits for their cell phone chip set, that involves designing matching circuits for their power circuits.

 

[yungman]

Want to get important knowledge in EE world, take a look at this book. It is quite easy, it has a way to explain grounding, ground current, EMI etc. in a very simple way with minimal electromagnetic theory. THIS is what is really important on the job.
https://www.amazon.com/dp/0133957241/?tag=pfamazon01-20

In today's high speed mixed signal environment, ADC, DAC, USB, 802.xx type of designs, knowledge of this is a MUST. You don't have to go into RF world to need this. Even if you are not planning to work as signal integrity engineer, you need this to design circuits and layout pcb. DON'T even think of depending on pcb layout designer to help you on this. It's on you to make the circuit work.

I don't even know whether people even hire pcb designers now a days, my last contract work I did a few years ago, we all design our own pcb. Good thing is I did my contract at home, barely went to the office once a week for an hour or so. To pass CE or UL, you better know signal integrity to pass those tests.

 

[jasonRF]

I've seen people from UC Berkley and Standford, they just so hung up with useless things, I would fired them if they were under me. RESULTS.

That's the problem, so many people studied and get good in all the formulas, but they don't have what it takes to design, getting the job done. That's the one I want to eliminate. Believe me, there are plenty of those people. Some people just don't have it, I just want to find people that have it.

Hi yungman,

Our paths haven't crossed for a number of years now - it is good to hear from you!

Clearly you worked at companies that expected new-hires right out of college to hit the ground running on circuit design. If there was a sufficient supply of candidates that met your requirements then it is easy to understand why you hired that way.

While I understand why you didn't hire those people, I think you are being pretty harsh by assuming that "they don't have what it takes", when in reality they may never have been taught to think about circuits this way and never read the kind of book that you mentioned reading to learn this stuff. This is really a failing of the EE curriculum. I know in my EE program there were only two courses in analog circuits and electronics that were required, and I don't recall either teaching this kind of thinking (or if they did, I just don't have what it takes ) . Instead, they taught basic semiconductor physics and attempted to give an intuitive understanding about how things worked at the device-level, and then moved on to hybrid-pi models and kvl/kcl... I think learning the intuitive understanding of circuits would have been a better choice, and I recall my advisor complaining about the way those courses were taught. Perhaps the upper-division courses on analog circuit design imparted this way of thinking, but I never took those courses. I was too busy taking extra courses on electromagnetics, microwave engineering, antennas, radar, physics and math. Today I can answer your transistor question only because a decade ago I got into hobby electronics for a short period of time; the op-amp question is something I could have answered by the end of my sophomore year.

By the way, I looked at Stanford's program and according to their web-site their EE program requires only one class each in analog and digital circuits
https://exploredegrees.stanford.edu/schoolofengineering/#majorstext
Of course they can take more if they want, but it is not required unless they pick particular specializations. The school I went to now has similar requirements as Stanford. These kinds of programs can be great at preparing students for graduate school and give students a lot of flexibility to prepare for a wide variety of careers. But some students (like me, and the ones you interviewed) don't leave with strong electronics skills - some because they didn't take the extra classes and some because they are better suited to other fields. On the other hand, there are a number of EE programs that require students to take several years of electronics classes; I suspect you had better luck hiring from those schools. I randomly picked Cal Poly to google and see that their well-regarded program is an example that fits in that category
https://ee.calpoly.edu/academics/undergraduate-program

cheers,

Jason

 

[berkeman]

Want to get important knowledge in EE world, take a look at this book. It is quite easy, it has a way to explain grounding, ground current, EMI etc. in a very simple way with minimal electromagnetic theory. THIS is what is really important on the job.
https://www.amazon.com/High-Speed-D...x=signal+int,aps,256&sr=8-3&tag=pfamazon01-20

I know that book quite well. Howie Johnson was one of the first engineers to work at the startup that I joined many years ago (now public and acquired several times by larger companies). After helping the startup to get going, he moved on and published that book (I believe he also wrote a regular column for EDN or some other big EE periodical). When he published his book, our VP of Engineering bought copies of the book for all of the engineers at the company, so we all ended up with it on our bookshelves. I did a lot of work in EMI and RF, so I referred to his book fairly often. It's one of the few books that I've kept over the years.

Small world!

 

[yungman]

I know that book quite well. Howie Johnson was one of the first engineers to work at the startup that I joined many years ago (now public and acquired several times by larger companies). After helping the startup to get going, he moved on and published that book (I believe he also wrote a regular column for EDN or some other big EE periodical). When he published his book, our VP of Engineering bought copies of the book for all of the engineers at the company, so we all ended up with it on our bookshelves. I did a lot of work in EMI and RF, so I referred to his book fairly often. It's one of the few books that I've kept over the years.

Small world!

This is actually about the most important knowledge an EE need to have in this mixed signal world. I don't think it's realistic to say "I just want to design digital or analog" in the industry. Everything is mixed, high frequency analog mixed with high speed digital circuit coexist in a tiny area of the pcb.

In 2000, when I was working for a small startup company, they actually had a specialist come in for 2 days to teach signal integrity. That's the first time I even heard of it. I was so into that, out of 6 EE, I was the only one that interact with the teacher while the other were kind of quiet...like dosing off. It was funny when I told my co-workers I really enjoy this, they said " you can have it all!". It's amazing I can predict how the current on the ground plane flows, that I can control them the way I want it.

Howard Johnson book is very good and easy. It's very common sense, minimal EM knowledge required even though it is related. Even though it's tittle is about digital, it's every bit applicable to analog design as antanna works both ways, the better it can transmit also mean the better it is to receive.

I still think EMI and RF is the most challenging and most fun area to work on. It's a totally different world. Not only you need a lot of knowledge, you need good pcb layout skill. I really enjoy pcb layout, that's where the magic happens. Now this is getting deep into electronics, very deep. To me, the most fun part is design with all distributed element using only microstrip to create capacitors, inductors but dancing around the Smith Chart. That the final pcb looks like a maze with pattern of copper.

 

[yungman]

Hi yungman,

Our paths haven't crossed for a number of years now - it is good to hear from you!

Clearly you worked at companies that expected new-hires right out of college to hit the ground running on circuit design. If there was a sufficient supply of candidates that met your requirements then it is easy to understand why you hired that way.

While I understand why you didn't hire those people, I think you are being pretty harsh by assuming that "they don't have what it takes", when in reality they may never have been taught to think about circuits this way and never read the kind of book that you mentioned reading to learn this stuff. This is really a failing of the EE curriculum. I know in my EE program there were only two courses in analog circuits and electronics that were required, and I don't recall either teaching this kind of thinking (or if they did, I just don't have what it takes ) . Instead, they taught basic semiconductor physics and attempted to give an intuitive understanding about how things worked at the device-level, and then moved on to hybrid-pi models and kvl/kcl... I think learning the intuitive understanding of circuits would have been a better choice, and I recall my advisor complaining about the way those courses were taught. Perhaps the upper-division courses on analog circuit design imparted this way of thinking, but I never took those courses. I was too busy taking extra courses on electromagnetics, microwave engineering, antennas, radar, physics and math. Today I can answer your transistor question only because a decade ago I got into hobby electronics for a short period of time; the op-amp question is something I could have answered by the end of my sophomore year.

By the way, I looked at Stanford's program and according to their web-site their EE program requires only one class each in analog and digital circuits
https://exploredegrees.stanford.edu/schoolofengineering/#majorstext
Of course they can take more if they want, but it is not required unless they pick particular specializations. The school I went to now has similar requirements as Stanford. These kinds of programs can be great at preparing students for graduate school and give students a lot of flexibility to prepare for a wide variety of careers. But some students (like me, and the ones you interviewed) don't leave with strong electronics skills - some because they didn't take the extra classes and some because they are better suited to other fields. On the other hand, there are a number of EE programs that require students to take several years of electronics classes; I suspect you had better luck hiring from those schools. I randomly picked Cal Poly to google and see that their well-regarded program is an example that fits in that category
https://ee.calpoly.edu/academics/undergraduate-program

cheers,

Jason

I have been away doing my own things for a few years. I was here when I was studying PDE and EM for the fun of it. I have been designing high end hifi power amps for a few years, I think I have hit the ceiling on that, so I start to pick up programming C++ and I am back here asking questions on that. It's just my crossword puzzle for old age to keep my mind from turning into jello! just keep learning from one thing to another.

I am absolutely surprised what you said about college EE major requirements. Only two classes of analog! I used to wonder what I missed not having an EE degree as my degree was biochem. Now I am grateful I did NOT take EE as my major! All these years, it's like learn when I need it. I kept changing jobs to different facet and study along the way to gain knowledge. From working for LeCroy designing ADC modules ( like digital scope), the Exar designing analog IC, to Seimens designing ultrasound imaging machine then to mass spectrometers then RF communication. Just kept changing fields and study along the way. So I just look at what is needed in the jobs.

What are they studying in the 4 years? All digital, embedded, firmware, FPGA etc.? In today's high speed mixed signal world, these are just part of the knowledge. Without high speed analog design, you won't go too far with just digital knowledge.

Semiconductor physics is good, but in my IC design job, I learned it from the older edition of this in 1984:
https://www.amazon.com/dp/8126521481/?tag=pfamazon01-20
It had a lot of circuit design, current mirror, differential pairs and all that. Anyone that took a course in semiconductor design should have plenty of circuit experience. I have to say, even in designing analog IC, I did not use those semiconductor physics, it's just "good to know". Actually design was just like normal circuit, you just use the available transistors, their limits. Using their resistors and get matching by layout etc. To me, it's just design with different available parts from discrete circuits. But of cause, I am talking about 30+ years ago, things might have been changed.

Yeh, I might be a little harsh, but the world is not like in college. I am sure now is good time as jobs are plenty. But believe me, I've seen market flipped around a few times. In good times, you have a pulse, you'll be hired. But in bad times, believe me, it is harsh. If you ever look for jobs between 2001 to 2005, you know what I mean. I had one interview on RF engineer position. I went two interviews, I got 100% on their test, that was on Smith Charts matching and all. I did NOT get the job because they said they need someone with more experience. That was harsh. My point is to make students marketable by learning the right thing even though it might sound harsh.

I worked for more than one startup companies, you get hired...and FIRED just as fast. I've seen people come and go in a month. That's harsh.

 

[Averagesupernova]

@yungman I really respect your high expectations. I feel as if in the past I have nearly forgotten that some of the formal laws in electronics exist such as KCL and KVL. It hit me one day when a coworker who troubleshot test equipment as I did asked me how I 'did it'. Apparently they thought I had some special magical view into what I was working on that they did not. Since this was a good friend I took a long time to give an answer that I thought accurately describes my approach at troubleshooting. What I learned by analyzing myself was that I was doing a lot of things that I was taking for granted. As a comparison, it was like describing every last little muscle movement involved in walking to someone who is learning to walk. I had no idea I was doing all of these things without thinking about it. You know why that happens? Practice practice practice practice. The only reason I may have been more special was because I probably enjoyed it more than anyone else I knew at the time.
-
So just because you think you aren't doing what someone less experienced than yourself does, doesn't mean it's true. You just aren't thinking about it.
-
And yes I looked at the circuits you posted. Simple stuff to anyone with experience.

 

[Jody]

I can't do that op-amp one in my head :( I thought it was obvious, but then I noticed the resistors weren't balanced and I can't remember the formula by heart anymore.